elementcollector1
International Hazard
Posts: 2684
Registered: 28-12-2011
Location: The Known Universe
Member Is Offline
Mood: Molten
|
|
Building a high-temperature induction furnace
I've decided to attempt to recreate the Czochralski process for making large gemstone boules. This process essentially goes as follows:
1) Melt a vat of very pure starting reagents.
2) Insert a seed crystal into the melt, which is attached to a threaded rod.
3) Rotate the threaded rod to draw the seed crystal upward, and rotate the crucible and melt in the opposite direction. If done right, a single
crystal boule should form around the seed and continue to be drawn until the seed reaches its maximum height.
4) Lift the newly-created gemstone boule out of the melt and turn off the heating.
Unfortunately, while I have the specifications for the rest of the procedure more or less worked out (using an Arduino, a low-RPM/high-torque motor
and a threaded rod held in place), I'm a bit stuck on how to accomplish the first step for sapphire (aluminum oxide). This has a melting point of
~2100 C, meaning conventional furnaces won't cut it - and neither will conventional insulation. I'm planning to use either rockwool or carbon foam for
the insulation, and a graphite crucible to hold the molten alumina, but I'm still a bit lost as to how to get such high temperatures for a relatively
large mass of alumina.
I turned to induction heaters as a means of heating that wouldn't damage the insulation (as oxyhydrogen might, among other problems) or be expensive
to source and maintain (as with oxyacetylene setups). While I understand the basics of the theory (that heating rate is proportional to the square of
the magnetic field produced for all materials, that different materials heat at different rates due to differing resistances, and that more power
input makes things heat faster), I'm a bit stuck on how I can apply this to building one.
I found this walkthrough that shows how to build a simple, low-power version (less than 1 kW), claiming derivation from the 'Royer oscillator' circuit (which
I'm unfamiliar with) which looks fairly simple to construct, but it also claims it cannot be scaled up well to higher masses or power input, as the
current draw for high-mass (and high-heat-capacity) materials becomes too great for the power supply to handle.
My main question is, how might I build an induction heater that is strong and robust enough to melt alumina? According to the basic calculations on
the website, to get this furnace up to temperature in something like 30 minutes with a 1 kg charge of material would require a power input (minus that
lost to the surroundings) of about 1 kW. I've seen power input estimates ranging from 1.5 kW in to 300 W out, to 300 W in and 200 W out - I suppose
that depends on efficiency. If I used wall voltage, I could theoretically get up to 1.8 kW in (likely much less), but all the induction heaters I've
seen use DC power supplies to power oscillating circuits.
Elements Collected:52/87
Latest Acquired: Cl
Next in Line: Nd
|
|
|
WGTR
National Hazard
Posts: 971
Registered: 29-9-2013
Location: Online
Member Is Offline
Mood: Outline
|
|
An incandescent light bulb can have a filament temperature above 2000°C, yet for some bulbs you can comfortably hold the glass envelope in your hand.
Think about that for a moment.
Some furnaces operate at "normal" temperatures of around 1000°C, and then there is a small hot zone in the middle of the furnace that operates at
much higher temperatures, that contains the sample. That way the insulation only has to handle the lower temperature.
An aluminum container can be polished to a mirror shine on the inside. A filament operating inside the container, in a vacuum, will radiate energy to
the aluminum walls, which will mostly (hopefully) be reflected back to the filament, keeping it from losing too much energy. Meanwhile, the aluminum
surface has so much more surface area than the filament, that any energy that gets absorbed will only raise the temperature of the container a small
amount. Anything enclosed by the filament can get very, very, hot.
|
|
|
elementcollector1
International Hazard
Posts: 2684
Registered: 28-12-2011
Location: The Known Universe
Member Is Offline
Mood: Molten
|
|
Yes, but I'm looking to create quite a large 2000-degree Celsius hot zone. I imagine even with the setup you describe, the maximum temperature of the
hot zone is still limited by the melting point of the filament itself - this is why I discarded more traditional furnace approaches like nichrome and
kanthal wires. The tungsten in incandescent bulbs would reach the appropriate temperature, but would require inert atmosphere or vacuum in order to
function.
[Edited on 8/29/2018 by elementcollector1]
Elements Collected:52/87
Latest Acquired: Cl
Next in Line: Nd
|
|
|
Heavy Walter
Hazard to Others
Posts: 127
Registered: 17-12-2015
Location: Argentina
Member Is Offline
Mood: No Mood
|
|
Hi elementcollector
You will need to protect your crucible in an inert atmosphere, like Argon.
What crucible volume are you planning to use?
|
|
|
elementcollector1
International Hazard
Posts: 2684
Registered: 28-12-2011
Location: The Known Universe
Member Is Offline
Mood: Molten
|
|
I plan to make boules of at least 6" height by 3" diameter, so assuming the boule expands all the way to the walls of the crucible, that would mean an
absolute minimum volume of about 700 mL. Best to aim for 1000 mL to be safe.
EDIT: An alternative approach I thought of would be to use something similar to Grant Thompson's Metal Melter (https://www.youtube.com/watch?v=d5pGN6pqkyY), except the terminals would be replaced by copper strips that run the vertical length of a crucible
(for more even heating) and the 'metal' in between them would be the graphite crucible itself. That said, I presume the graphite would need to be
under vacuum at such temperatures to avoid catching on fire or otherwise destroying itself.
This might be a preferential approach for me, as I have already successfully built a spot welder (affectionately named 'Sparky') using his
instructions and own a vacuum pump. However, I am unsure if this method achieves the required temperature. Without resistance measurements of a
suitable graphite crucible to use, I have no way of knowing how long it will take to heat to a given temperature, nor at what temperature it will stop
heating.
[Edited on 8/29/2018 by elementcollector1]
Elements Collected:52/87
Latest Acquired: Cl
Next in Line: Nd
|
|
|
WGTR
National Hazard
Posts: 971
Registered: 29-9-2013
Location: Online
Member Is Offline
Mood: Outline
|
|
Graphite will readily react with oxygen at high temperatures, slowly (or quickly at 2000°C) burning away to carbon dioxide. An inert gas or vacuum
is needed for graphite. Vacuum will make it so much easier to heat things, as you wouldn't have the convective heat losses.
I posted some information for a small inductive heater here some time back. It's nowhere near the amount of power that you want, but maybe it contains some useful ideas. People are conditioned to not
mix electricity and water, but if the water is deionized and the circuit board is very clean (all flux, etc., removed with solvents), then it will
keep everything very cool without shorting anything out. At worst the water will boil at 100°C. Here I even have the working coil submerged in
water with the driver. Instead of using tubing or a solid wire for the working coil, I used Litz wire. This helps keep the AC losses low in the
coil, as you may have 100's of amps circulating in the tank circuit.
The Royer/Mazzilli circuits are very simple and prone to blowing up MOSFETs. It's better to AC couple the gates and bias the MOSFETs slightly "OFF"
instead of using the usual DC coupling. That way the circuit will not be self-starting, but it will also not blow up its MOSFETs either. To start
it, a momentary pushbutton can be used to pull the MOSFET gates up into the linear mode of the MOSFET, allowing it to start. While oscillating, the
gates will have the proper drive for the circuit to function. If the circuit quits oscillating (too much load, etc), then the gate bias falls off,
and the MOSFETs shut down, protecting them.
[Edited on 8-29-2018 by WGTR]
|
|
|
elementcollector1
International Hazard
Posts: 2684
Registered: 28-12-2011
Location: The Known Universe
Member Is Offline
Mood: Molten
|
|
That's a fascinating design - unfortunately, I don't have a sister. I may have to learn this strange 'braiding' technique on my own. I do happen to
have some 32 AWG magnet wire sitting in storage - maybe I'll give that coil design a try.
Encasing the whole apparatus in vacuum will be a bit harder, but if the graphite crucible is the only thing producing heat, I might be able to get
away with a lesser insulation. Still, I have most of the apparatus on hand already - the only thing left to do would be to try it.
Elements Collected:52/87
Latest Acquired: Cl
Next in Line: Nd
|
|
|
WGTR
National Hazard
Posts: 971
Registered: 29-9-2013
Location: Online
Member Is Offline
Mood: Outline
|
|
I think what you're planning to do is very difficult, but will be possible with a lot of planning and thought. The temperatures are very high and
will probably require a massive amount of power. Sounds like an interesting project though.
|
|
|
wg48
National Hazard
Posts: 821
Registered: 21-11-2015
Member Is Offline
Mood: No Mood
|
|
Ouch that’s a lot of things to figure out. I will start at the input power.
Cooker sockets in Europe and America are about 10kW. I know someone with a 3kW kiln that can reach 1,200C perhaps 1300C and it has a capacity of
about 2cuft. It table top sized, about the size of a large microwave oven. So 1Kg of Al2O3 in a graphite cruciple would fit in easily. Extrapolating
to 2,000C about 10kW would be required depending on the thermal mass of the insulation, its efficiency(heat loss) and the efficiency of the induction
heater.
Building a 10kW RF generator is a major task requiring significant knowledge, test gear and money. It has been done by lots of radio hams and a few
tesla coil builders.
Typical single platter induction hobs are about 2kW and cost about £50. Extrapolating to 10kW, that about £250. However they operate at typically
at 20kHz as they use magnetic hysterisis losses for generating the heat. I have not done the calculation but I would expect your graphite crucible to
require higher frequency for a practical solution as it uses resistive losses for heating. Probably in the 100kHz to Mhz range which makes the
components much more expensive say x10. That’s £2500 just for the heating power.
Perhaps an alternative is buy a used RF generator. I spotted an RF plastic welding machine on Ebay for £99 starting bid. Its probably in the kW range
of power but it may be three phase.
An other alternative is graphite heating elements (cheap graphite gouging rods) or the more expensive silicon carbide or molybdenum silicide heating
elements. The require a large transformer for the current and ideally a large variac. A bank of rewired microwave oven transformers would be very
cheap if not free and does not need serious electronics knowledge.
PS: Reading the other responses. For carbon heating elements and carbon foam insulation you can use nitrogen or argon so the enclosure does not need
to withstand a vacuum.
i
[Edited on 29-8-2018 by wg48]
Borosilicate glass:
Good temperature resistance and good thermal shock resistance but finite.
For normal, standard service typically 200-230°C, for short-term (minutes) service max 400°C
Maximum thermal shock resistance is 160°C
|
|
|
wg48
National Hazard
Posts: 821
Registered: 21-11-2015
Member Is Offline
Mood: No Mood
|
|
Apparently high frequency (>100kHz) is not needed to induction heat graphite. Probably due to its much high electrical resistance compared to
metals.
From a note on induction heating of graphite crucible it used 1kHz,5kHz and 9kHz using a water cooled 9 turn 16mm diameter copper pipe in a coil;
of 230mm diameter and 200mm long with a current 700A. That is in range of frequency and currents of induction heating hobs. The voltage or power was
not given. I assume similar to an induction hob, probably no more than a several kW at several hundred volt. That makes the semiconductors and
capacitors relative cheap compared RF components.
It may be possible to induction heat a 50ml crucible using the electronics of a induction hob driving a suitable copper pipe coil that has a similar
inductance as the pancake coil of the hob it replaces.
Attachment: graphite-indheat.pdf (443kB)
This file has been downloaded 614 times
Borosilicate glass:
Good temperature resistance and good thermal shock resistance but finite.
For normal, standard service typically 200-230°C, for short-term (minutes) service max 400°C
Maximum thermal shock resistance is 160°C
|
|
|
Texium
|
Thread Moved
27-11-2023 at 12:26 |
![[*]](images/xpblue/default_icon.gif){kind=icon}
