Professional Documents
Culture Documents
In an Electrolyte
Second Semester Report
Spring Semester 2007
By
Robert Allard
David Harris
Joe Koonce
Dan Roarty
__________________________________
Senior Design Coordinator
Abstract
Inductively coupled gas plasmas have been studied extensively and are used
less common subjects of study and applications appear to be very limited at this time.
This may arise from the difficulty in creating a plasma in this medium. Nevertheless, the
physical processes by which a plasma can be created in a liquid are similar to those
necessary to ignite and sustain one in a gas. This paper describes progress made toward
Air core inductors of various configurations and size were wound, measured and
tested using a 13.56 MHz RF power supply. Impedance matching was obtained with a
A gas plasma (helium at atmospheric pressure) was ignited and sustained using a
power. Attempts to establish a plasma in normal saline solution have been unsuccessful
so far. Each attempt under varying conditions resulted in partial vaporization of the
solution before any photon emission occurred. A glow at the perimeter of the meniscus
of solution remaining after partial vaporization was observed. Further work remains to
2
Table of Contents
Title
Abstract 2
Table of Contents 3
I. Introduction 4
B. Inductor Modeling 6
C. Matching Network 8
D. Experimental Results 9
E. Project Management 12
F. Budget 13
Bibliography 14
Appendix A – Photos 15
3
I. Introduction
design project in the typical sense of designing and building a device that performs some
useful function and might be marketable. It is rather more like a research project. In fact
first. For example, we thought that we could quantify the magnitude of the magnetic
field necessary to induce a plasma in the electrolyte and thus find an inductance value
and power supply specification sufficient to realize that field, much like working
backwards in a gas laser from output power to gain coefficient to population inversion to
rate equations to pump rate to input power, etc. That approach seemed sensible given our
undergraduate experience with “ideal” systems and analytic solutions to every problem.
understood that our undergraduate toolkit was inadequate to the task (see for example
Garrett). The real world after all is neither ideal nor tractably analytic.
us from wandering to far in the wrong direction. That didn’t happen either. In the end
we were able to test the possibility of a plasma in an electrolyte and though we didn’t
includes a brief review of previous work, inductor modeling, matching network solution,
and plasma tests and results. Part III presents conclusions about the work and a
4
II. Experiments and Results
The first semester of the project was consumed with much trial and more error as
we spent time trying a number of inductor configurations and materials with different RF
network for each inductor-power supply combination proved to be both challenging and
time consuming, in part because of our trial and error approach to the experiment. These
issues are likely trivial for an experienced engineer but they weren’t for us. As the
semester ended we identified a number of tasks to complete in the second phase of the
of a plasma in electrolyte,
software, and
approach.
Some of these tasks were accomplished, one was discarded and one still needs further
work. The details of our efforts follow in the remainder of the report.
5
B. Inductor Modeling
inductors in software and continuing the effort to create a plasma in the Advanced Energy
system from the first semester. We soon abandoned the plasma chamber idea, however
inductor modeling produced results. The intent of modeling was to determine the
inductor specifications that would produce the largest magnetic field in our liquid sample.
Geometry, number of turns, type and size of wire were all important factors. We used
Ansoft Maxwell 3D software to model magnetic flux density for different geometries.
The results are shown in the following table and in the examples of Maxwell output
which follow.
Configuration
B Field # # ID OD Length Spacing Inductance
(Gauss) Flux (W) Turns Layers Axial Radial (mm) (mm) (mm) (mm) (H) Notes
10 5.70E-08 2 1 x 5 9 6 2 5.70E-09 Cu
10 4.80E-07 3 1 x 5 9 10 2 4.80E-08 Cu
23 6.50E-07 3 1 x 5 9 6 0 6.50E-08 Cu
32.5 9.20E-07 6 1 x 5 9 12 0 9.20E-08 Cu
14 9.10E-07 2 2 x 5 15 2 1 9.10E-08 Cu
23 1.70E-06 4 4 x 5 27 2 1 1.70E-07 Cu
66.3 1.50E-05 36 6 x 4 72 34 2 1.50E-06 Cu
42.4 7.80E-06 18 6 x 4 72 16 2 7.80E-07 Cu
42.5 7.80E-06 18 6 x 4 72 16 2 7.80E-07 Tungsten
42.4 7.80E-06 18 6 x 4 72 16 2 7.80E-07 Aluminum
17 9.10E-07 3 3 x 4 36 2 2 9.10E-08 Cu wire dia = 4mm
6
The results of these simulations (and advice from our advisor) caused us to change our
thinking about inductor design. We moved toward smaller inductors, i.e. fewer number
7
C. Matching Network
Early on two models were created in Mathcad for series shunt and series Z impedance
matching, working with ideal devices and many assumptions. It was soon evident that the
modeling left out some parameters but at least got us near the ball park and in one case on
the infield. Thereafter, each attempt to create a match to the generator 50 ohm output
began with finding the load impedance using a network analyzer, then calculating the
value of capacitive and inductive components required to match this load to the generator
using paper Smith charts and WinSmith software. Once the correct component values
were determined, an L-match was assembled using fixed value capacitors and/or
inductors. On all but one occasion our match proved to be incorrect. This forced us to
return to our calculations to find the source of our error. We learned after some work
with WinSmith that the adequacy of the match was susceptible to very small variations (<
2 pF) in one or both of the component values. The obvious conclusion was that fixed
component values for the L match wouldn’t get us very far since the stated values of the
components were only nominal. We needed the ability to tune the match network. So we
scavenged two variable vacuum capacitors and “doorknob” capacitors and an enclosure
and built a standard L-match whose tuning range would fit our expected needs. The
20-500p
Source Load
7-1000p 700p
8
Tests were conducted to determine if the matchbox and various inductive loads
could be tuned to 50 ohms. Using an Advanced Energy Apex 10 kW, 13.56 MHz power
supply we initially obtained 47% forward power. Still this amount of power was enough
to raise inductor temperatures to well over 100º C in a short amount of time. We found
that the fixed capacitors, while getting us close to the desired range, reduced the effect of
the tunable capacitors so they were removed. We were then able to realize forward
D. Experimental Results
Though we discarded the idea of using the AE plasma chamber, we retained the
approach of first attempting to create a plasma in a gas before working with solutions.
Gas plasmas are well understood and Dr. Collins’ lab has long experience with them. We
began by making a variety of inductors using 14 ga magnet wire. A jig was created from
supply. The pulsed supply would yield a higher di/dt and bigger electric field according
DC supply (constant output), power FETs, power resistors, a function generator and a
small Hewlett Packard DC power supply. The function generator drove the FET gate at
the desired frequency and waveform and thus switched the DC ouput from the Pinnacle
to produce the desired pulsed power. After much tweaking and modification the system
worked reasonably well at low frequency (e.g. 1 kHz). The output as measured by an
oscilloscope was a recognizable DC pulse at the expected frequency with some noise and
9
ringing at pulse rise. Some of the noise was reduced by tying all components to common
ground.
We assumed that we would need to operate the supply at 100 kHz minimum (this
conducted the previous year on another project in the lab). However, when the gate drive
was increased above 10 kHz the output was no longer recognizable as a square wave.
Tests demonstrated that the function generator and DC supplies were working correctly
so we surmised that the power FET’s parasitic capacitance was too high to permit
switching at the desired frequency. The pulsed power supply was inadequate so we
wound and were designed to enclose a Pyrex glass tube of 2 mm approximate ID and 30
cm approximate length with appropriate hardware for gas flow. Some of the air core
inductors made were 6, 12 and 46 turn coil, single layer; 24 and 40 turn coil, double
The first inductor used was the 12 turn single layer coil. The coil was soldered
reactance was .125 + j7.24 or 84 nH at 13.56 MHz. The glass tube apparatus and
inductor were set up, helium was introduced and the power supply energized at 100 W
(the minimum available from the AE Apex). Forward power was >90% but no plasma
was ignited. We changed to the 46 turn coil (impedance of .24 + j74.13, 870 nH), and
repeated the test at the same gas flow rate and power but were unsuccessful again. At
this point we decided that some type of “spark” might be needed for ignition. This
10
technique had been used previously in the lab in other experiments. So we simply placed
a conductor (a wire about 100mm long) into the glass tube figuring that the electric field
inside the tube might create an arc between wire and wall. It worked; an atmospheric gas
plasma was ignited and sustained after the wire was removed. Photos of the apparatus
inductance, and lowest power, required to ignite the plasma. An Advanced Energy RF Z-
Scan was used to facilitate tuning of the match network for each trial. We were unable to
ignite plasmas with either a 6 turn coil or the 9 and 13 turn planar inductors or the double
layer coils.
At this point we changed power supplies since the 100 W lower limit of the Apex
was more power than was needed or desirable. We used an Advanced Energy RFG 1250
2 kW 13.56 MHz supply. A gas plasma was again ignited using the 46 turn coil at a
with a computer interface as does the Apex so we were unable to record forward,
reflected and load power easily. The RFG 1250 uses an analog meter to display voltage
but at the low power level we were using (judging by the small excursion of the variable
voltage control) the meter reading was not meaningful. We certainly could have used test
equipment to measure delivered power but that wasn’t particularly valuable information
at that point.
Having produced a plasma in gas, we proceeded to the real aim of the project –
plasma in a liquid. The Pyrex tube was modified to contain a liquid. 0.9% NaCl (normal
saline) was placed in the tube and power was applied using the 46 turn inductor. As
11
power was increased the expected happened – the liquid began to boil in a matter of
seconds and exited the tube with considerable velocity. We varied the rate at which
power was increased to the load for several trials but the results were identical to the first
attempt.
Further tests were conducted using a planar coil but this configuration didn’t
work; the inductor just overheated. We then tried pulsing power to the load by setting the
power (voltage) control at a variety of levels substantially above minimum power and
then manually switching the RF on and off. It was thought that increasing di/dt in this
processes kicked in. Vaporization and ejection of the liquid was again the result.
However, on two of these attempts we did observe a short-lived yellow glow at the
perimeter of the meniscus of the liquid remaining in the glass tube which was now a few
millimeters below the lower end of the inductor. Our advisor observed this event also
but, like us, was unsure of its nature and origin. He suggested we use a high speed
camera to record the event for further analysis. At the time of this report we have not
done this.
The results of these trials with liquid make it abundantly clear that we will need to
find a means to confine the liquid as power is applied to the inductor. Obviously as the
liquid temperature increases so will pressure and a vessel that can withstand high
E. Project Management
and held to discuss theory, experimental setups and results. Weekly project updates were
12
required of all team members. Meetings were normally held on Wednesdays from 1:00 –
2:30 pm. Lab times were set for at least 2 hours starting at 1:00 pm on Mondays and
Fridays. Weekly minutes were distributed to all team members, our advisors, and
instructors. Meeting attendance was reported in the minutes as well as weekly progress
reports which also included discussions of experiment design, setup, and expectations.
F. Budget
As we explained in the first semester’s report, we did not establish a budget for
this project and in fact have spent little money to this point. We purchased power
resistors from a local vendor for $5.76. All other materials and equipment we needed
liquid occurred even at relatively low power with every trial. As noted earlier, the liquid
should be quantified and then materials found that will tolerate that regime. A quartz
tube is one possibility that could be explored. It would also be interesting to image the
13
Bibliography
B. N. Chapman, “Glow Discharge Processes”, New York: John Wiley and Sons, 1980.
Fields, Part I”, Journal of Applied Physics, vol. 22, No.9, pp. 1091-1107, Sept. 1951.
for an Inductively Coupled RF Plasma Source”, Plasma Sources Sci. Technol., vol. 7, pp.
179-185, 1998.
Coupled Plasma Sources, Journal of Applied Physics, Vol. 85, No.3, pp. 1366-1371, Feb.
1999.
14
Appendix A – Photographs
15
Acknowledgements
The team members wish to express our appreciation and gratitude to Dr. Cameron Moore
and Dr. George Collins for their assistance, encouragement and technical advice. On
multiple occasions we were mired in quick sand and Cam threw us a life line. His
16