Slumping Splits

 This is a description of the analysis process to determine the possible causes of a split during a slump.     

Credit: Maureen Nolan

Observe the piece.

It is a tack fused piece, about 20cm (8") square, which has been slumped. 

The base layer is of clear. The piece has three additional layers, but the fourth layer is only of small glass dots and rectangles.  The central, heart, area is made of three layers.

A split has appeared during the slump. It is split irregularly through pieces rather than around them.  It is split through the thickness but only partially across the piece.

In one area the (brown) third of four layers spans the split.  Further to the left a brown second layer seems to have broken, but still spans the split.

Threads and particles of glass are connecting across the split. 

The edges are probably sharp, although only so much can be deduced from a description and one photograph.

History of the Piece

The tack fused piece has been put in a mould to form a platter and has split during the slump.

The schedule in essence was:

139ºC/250ºF to 565ºC/1050ºF for a 30’ soak (some pauses but all at a ramp rate of 139ºC/250ºF)

83ºC/150ºF to 688ºC/1270ºF for 10’

222ºC/400ºF to 516ºC /960ºF for 60’

111ºC/200ºF to 427ºC/800ºF for 10’

167ºC/300ºF to 38ºC/100ºF, off

 

The assumption is that the tack fused piece received a similar annealing soak and cool.

 

Diagnosis

Too fast

Slumping a tack fused piece of three layers plus decorative elements on top needs to be fired as for 19mm (6 layers) minimum (twice the actual).  My work for the Low Temperature Kilnforming* eBook showed best results are achieved by slumping as for one more layer (21 mm/0.825” in this case).  This gives a proposed schedule of:  

120ºC/216ºF to 630ºC/1166ºF (not 688ºC/1270ºF) but for 30 to 45 minutes

AFAP (not 400ºF) to anneal 516ºC/960ºF for 3.5 hours (not 1 hour)

20ºC/36ºF to 427ºC/800ºF, 0

36ºC/65ºF to 371ºC/700ºF,0

120ºC/216ºF to room temperature

 
Commentary on the proposed schedule:

The slump is relatively shallow, so a low temperature with a long soak is the most suitable schedule for this piece.  The drop to anneal is at a sedate rate of 222ºC/400ºF.  This is inappropriate, generally.  Just as there is a rapid rate to top temperature to avoid devitrification, so there needs to be an AFAP drop to anneal, also to avoid devitrification.  The anneal soak was not the cause of the break, but it is worthwhile noting the recommended anneal soak and cool rates are longer and slower than that used.  This is a note for the future.

 

Suspect Tack Fuse

If the tack fuse schedule was like the slump schedule, the slump was started with inadequate annealing in the previous firing.  More importantly, the evidence for an inadequate tack fuse is that the split under the brown rectangle was through the clear and red on top, but the split left the brown intact.  This means it was not securely fixed to the red below it. 

 

If the condition of the tack fuse is not sound, it is probable that difficulties will be experienced in the slump.  The poster commented “… why do [these splits] happen only when slumping – it came through tack just fine.”    It is probable the tack fuse was not “just fine”.  The way to be sure the previous firing was just fine, is to test for stress.

 

There is enough clear in this piece that an inspection for stress could be conducted by use of polarising filters before the slump.  Testing for stress is a simple viewing of the piece between two sheets of polarised light filters.  Doing this test will give information on the amount of stress, if any, in the flat tack fused blank.

 

Slump Split

During slumping the glass is subjected to more movement and therefore stress than while being fired flat.  The glass is often only barely out of the brittle zone when being slumped and that makes the stress more evident during the early part of the slump. This requires careful inspection of the failed piece.

 

Look at the glass surrounding the split.  My opinion is that the edges are sharp.  If rounded, the threads of glass from the edges of white would have melted to the edges of the split rather than spanning it. 

 

It appears the top layers were hot enough for less viscous glass on top to form stringers that span the break as the underlying layers split.  It is probable that the split was during the plastic phase of the slump for the upper glass, but  the lower layers were not as hot and suffered thermal shock. 

 

This split of lower layers, while the overlying ones are whole, is often seen in tack fuses, although the top ones do slump into the gap as the firing proceeds.  In a slump there is not enough heat, time or space, for the brown piece to slump into the gap.  Both splits appear to be a result of too rapid firing.  In the flat fusing work, the split results from too fast a ramp rate during the brittle phase of the glass.  But the slumping splits appear to occur after the brittle phase, almost as a slow tear in the glass. This may result from the differential heating of the layers if not fully combined.  It may also indicate the split developed slowly. 

 

One other observation is that these splits seem to be more frequent during the slumping of tack fused pieces.  As speculated above, it may be the inadequate tacking together of the pieces of glass during the first firing, which can form a discontinuity in transmitting heat.  And it may be that the different thicknesses across the tack fused piece allow stress to build from differential heating of the glass.

 

Rates

 

Whichever of these speculative effects may be true, it appears the ramp rates are suspect.  As mentioned elsewhere* (and in Kilnforming Principles and Practice to be published soon), the reasons for these splits are not fully known.  Even microscopic examination by Ted Sawyer has not produced a satisfactory explanation.  The only practical approach that has been successful is to slow the ramp rates.  However, the appearance of these splits is essentially random (with our current understanding), so prevention is difficult.

 

Conclusion

The probable cause of the split in the slump has been that the ramp rates were too fast.  This may have been made worse by the too short anneal soak, and the too fast cool of the tack fused blank.

 

Remedy

There is no practical rescue for this piece.  Prevention in the future is to use ramp rates that are for at least one layer thicker, if it is full fused.  If it is tack fused, firing as for twice the thickest part plus one additional layer is advisable to slow the ramp rates, allowing all the glass to heat and form at the same rate.

 

 

*Low Temperature Kilnforming; an Evidence-Based Approach to Scheduling.  Available from:

Bullseye

and

Etsy

Sample Tiles

credit: Tia Murphy

There are advocates for making tiles as references for future work.  

• They show the profiles achieved at different temperatures.  
• They can be stored for easy visual reference when planning a firing.  
• It is a useful practice for any kiln new to the user.  

These tiles are assembled in identical ways to enable comparisons.  They should include black and white, iridised pieces- up and down, transparent and opal, and optionally stringers, confetti, millefiori, frit and enamels.  

The tiles are fired at different top temperatures with the same heat up schedule with the top temperature of each at about 10C or 20F intervals.  These show what effect different temperatures give.  Start the temperature intervals at about 720C or 1330F.

This is a good practice, even if time consuming.  It gets you familiar with your kiln and its operation.  It gives a reference for the profiles that are achieved with different temperatures at the rates used.

Ramp rate and time

But, as with many things in kilnforming, it is a little more complicated.  The effect you achieve is affected by rate and time used as well as the temperature.

The firing rate is almost as important as the temperature.  

• A slow rate to the same top temperature will give a different result than a fast rate.  
• The amount of heat work put into the glass will affect the temperature required.  
• Slow rates increase the time available for the glass to absorb the heat.  
• Glass absorbs heat slowly, so the longer the time used by slower rates, the rounder the profile will be.

Since time is a significant factor in achieving a given profile, any soaks/holds in the schedule will affect the profile at a set temperature.  A schedule without a bubble squeeze will give a different result than one with a bubble squeeze at the same temperature.

To help achieve knowledge of the rate/time effect, make some further test tiles.  Use different rates and soaks for the test tiles of the same nature as the first temperature tests. But vary only one of those factors at a time. Consider the results of these tests when writing the schedule for more complex or thicker layups. 

Mass

Also be aware that more mass takes longer to achieve the same profile.  Slower rates and longer times will help to achieve the desired profile at a lower temperature.  It is probably not practical to make a whole series of test tiles for thicker items.  But, a sample or two of different thicknesses and mass will be helpful to give a guide to the amount of adjustment required to achieve the desired outcome.

The results of sample tiles are due to more than just temperature.  They are a combination of rate, time, and temperature (and sometimes mass).  These factors need to be considered when devising or evaluating a schedule, because without considering those factors, it is not possible to accurately evaluate the relevance of a suggested top temperature.

See also: Low Temperature Kilnforming, available from Bullseye and Etsy

Scheduling for Thick Landscapes

Thick slabs often involve numerous firings of increasingly thick work.  I am using an existing example, with their permission, of the first stages of a thick landscape.  The initial concern was with bubbles in the first layup, then the strategy for firing the thick slab.

Plans

This is the first part of a landscape with depth.  It will be fired 5-7 more times.  This first piece will be inverted for the next firing with the clear facing up, to avoid reactions between the colours.  It is similar to an open face casting. There is a Bullseye Tip Sheet on open face casting that will give a lot of information.

Layup

Picture credit: Osnat Menshes

This work has a base of clear that is mostly overlaid with one layer of 3mm pieces, although in some places another layer, and there are some pre-fired elements as well.  It is fired on Thinfire shelf paper.

Bubbles 

There is concern about the number and size of the bubbles after the firing, and how to avoid them.  Will they grow over the multiple firings?

The many small bubbles are characteristic of kilnformed glass.  The few larger bubbles may result from the frit that is under the pieces that form the top surface.  And there are some overlaps of clear over colour that may form pockets where air can collect. I advise leaving the scattering of the frit until all the decorative pieces are in place.  The bubbles will migrate toward the top during the multiple firings.  They will not grow in size unless they combine during the upward migration.  A later suggestion about reducing the number of firings will reduce the bubble migration and risk of increasing in size.

Picture credit: Osnat Menshes

Schedule

Proposed Schedule (Temperatures in degrees Celsius)

1: 180 – 560, 30’    I would go to 610 for 30'

2: 25 – 680, 120’    I would use only 30'

3: 220 – 810, 15’    I would set the top temperature at 816, 15’.

4: 9999 – 593, 30’  Eliminate this segment. 

5: 9999 – 482, 120’ I suggest one hour soak

8: 55 – 370, off      83 – 427, 0’

7: 150 – 371, 0’

8: 330 – to room temperature, off.

 

Eliminate segment number 4.  Any temperature equalisation done at this temperature, is undone by the AFAP to the  annealing.  The temperature equalisation occurs at the annealing temperature. No soak at an intermediate temperature is required.  This blog post gives some information about annealing above and below the annealing point (Tg). 

Firing Incremental Layers

The plan is for five to seven more firings.  Continuing to build up the thickness on each firing, may have some problems.

• There is increased risk of compatibility problems when firing a piece to full fuse many times.
• There is a risk of more bubbles and of the existing ones becoming larger as they move upwards and combine with other smaller ones.
• With each firing the thickness is increasing and so becoming a longer firing.  This is because the heat up, annealing, and cooling each need to be longer.  For example - 6mm needs 3hour cooling, 12mm needs 5 hours, 19mm needs 9 hours. 

Multiple Slabs

These are the main reasons that I recommend firing a series of 6mm slabs separately and combining them in one final firing.  Firing a series of 6mm slabs and then combining them in a single long and slow final firing has advantages.

• The individual pieces do not need to go through so many full fuse firings, reducing the risk of compatibility problems.
• The small bubbles in each firing will not have the chance to rise through all the layers to become larger.
• The total time in the kiln for the combined pieces will be less than adding layers to already fired layers.

Examples

It is often difficult to convince people that firing by adding incrementally to an existing slab, longer firing times are required than by firing a group of 6mm slabs and a single combined firing of all the slabs.  I give an example to illustrate the differences.

Annealing

Assume there are to be a total of eight firings (existing 6mm slab and 3mm for each of seven more firings).  Also assume that each additional firing is of 3mm. This makes a total of 28mm.  Compare annealing and cooling times for each firing:

Firing      thickness       anneal and cool (hours minimum)

1            6mm                    3

2            9mm                    4

3            12mm                   5

4            15mm                   7

5            18mm                   9

6            21mm                   11.5

7            25mm                   14

8            28mm                   17

Total                                   70.5 hours annealing time (minimum)

To fire up 5 six millimetre slabs takes less time – 3 hours annealing and cooling time for each firing cumulates to 15 hours.  Add to that the final firing of 17 hours annealing time.  A total of 32 hours.  This is half the time of adding to the existing slab at each firing.  Multiple 6mm slabs can be fired at the one time if there is space in the kiln, which would reduce the kiln time for the 6mm slabs even further. 

An additional advantage of firing 6mm slabs and combining them, is that bubbles can be squeezed out more easily in the final thick slab fring because of the combined weight of the  slabs.  You could make the individual slabs a little thicker, but that would involve damming each slab.  Not an impossible task of course.  And it would change the calculations, by reducing the number of firings.

Heat Up

Another time saving is to use the second cooling rate from the Bullseye document Annealing Thick Slabs as the first up ramp rate. Take this rate up to a minimum of 540˚C. Although, this is an arbitrary temperature above the strain point to ensure all the glass is above the brittle phase.  It is possible to maintain this initial rate to the bubble squeeze.  But with the slow rises in temperature required for thicker slabs, it is sensible to increase the rate from 540 to bubble squeeze to reduce the firing time.  Once past the bubble squeeze a more rapid rate can be used to the top temperature.  

The heat up times could be about half the minimum cooling times.

A worked example (with certain assumptions) would be:

Firing      thickness       time to top temperature total time.

1            6mm             6.3               

2            9mm             7.1

3            12mm            8.4

4            15mm            10.7

5            18mm            15.9

6            21mm            19.4

7            25mm            25.1

8            28mm            29.1              ca.122 hours

But firing five times for 6mm equals 31.5 hours plus the final firing up of 29.1 hours equals a total of 60.6 hours.  Again about one half the time of progressively building up a base slab to the final thickness.

Savings

This example shows that approximately 90 hours of firing time can be saved by making a series of six millimetre slabs and combining them in a final firing.  There is the additional advantage of reducing the occurrence of bubbles between the layers in the final firing because of the weight of the combined slabs.

Heat Up Soaks

Photo credit: Bullseye Glass Co.

It is often advocated that there should be a soak at the strain point to even out the temperature throughout the glass.

My question continues to be why? 

The glass has survived whatever rate has been used up to that point during its brittle phase.  So, it already has every chance of surviving a rapid rate during the plastic phase.

Instead of a soak at the strain point, Bob Leatherbarrow indicates a soak during the brittle phase will be more successful in avoiding heat up breaks.  He has observed that heat up breaks are most likely to happen around 260ºC/500ºF.  Therefore, a soak in that region is most likely to be of use in evenly distributing the heat effectively through the glass rather than at a higher temperature.  He recommends up to a half hour soak there before proceeding at the same rate to the strain point (about 540ºC/1004ºF).  The ramp rate to this heat up soak in the brittle phase should be related to the thickness of the glass and the intended profile.

The thickness to be fired for is determined by the profile.  Rates for full and contour fusing can be as for the thickness before firing.  Rounded tack fuse needs to be fired as though twice as thick, and sharp tack or laminated fuse need to be fired as though 2.5 times.  More information on initial ramp rates to the strain point can be found in Low Temperature Kilnforming available from Bullseye and from Etsy

Deep Slumps with Bubbles

 

Photo Credit: Rachel Meadows-Ibrahim

The main causes of the large thin bubble is most probably  too high a temperature combined with a long soak.

Elevation of the Mould

The poster indicated there are eight holes total – four on the sides and four under the glass. This means any air has an exit out from under the glass and from the inside of the mould. So, in this case it does not need to be elevated for exit of air.  In my practice l have never, except in tests, elevated my slumping moulds. I have not had failures. My experiments involved in writing the eBook Low Temperature Kilnforming  showed no significant temperature differences between elevating, or not, below the mould.

Effect of Fast Rates

Slow rates to low temperatures with long soaks avoid sealing the glass to the mould. This means air can move out from under the glass during the slump. 

• Fast rates, and elevated temperatures can restrict air movement from under the slumping glass.  
• Fast rates and high slump temperatures can each cause uprisings because the glass slides down the mould during the soak, and that weight pushes the bottom upwards.

Temperature and Uprisings

This uprising is different from the bubble at the bottom on this piece. It is possible to see the glass bubble is thinner than the surrounding glass. As there were holes for air to escape, it seems the temperature and or speed was great enough to allow the glass to form to the mould at the bottom.  This covered the air holes and allowed the remaining air to push upwards on the glass.  A lower top temperature may have avoided this bubble formation.  Certainly, a combination of a slower rate and a lower temperature would have avoided the formation of the bubble.

Observation

Further, observation during the firing would have caught this bubble formation early enough to skip to the annealing and result in a piece with only a slight uprising, and before it became a bubble. Peeking should start at the beginning of the slumping soak and be repeated at 5 to 10 minute intervals.

Visible Devitrification

"Why does devitrification appear on slumped pieces?"

A brief explanation 

Scientific research in developing a glass matrix to support bone grafts gives some information.  This kind of glass matrix requires to be strong.  Development showed that devitrification weakens the matrix.  The crystals in a matrix are not as strong as the amorphous glassy state.  So, devitrification needs to be avoided.

The research to avoid devitrification showed that it begins at about 600˚C/1110˚F.  It only begins to become visible above 700˚C/1300˚F.  The process developed was to introduce a “foaming” agent.  The process fired slowly to 600˚C/1110 ˚F and then quickly to 830˚C/ 1530˚F.  It left a strong open matrix around which bone can grow. Although the research used float glass, it is also a soda lime glass, just as fusing glasses are.  The formation of devitrification begins at the same temperature for fusing glasses as for float.

The result of this medical research shows that devitrification begins on glass before it is visible. Devitrification is cumulative. A little becomes greater with another firing.  This is so even with good cleaning between firings. The new devitrification builds on the previous.  It does this from 600˚C/1110 ˚F.

A subsequent firing can continue this devitrification to the point where it is visible. This can happen, although the temperature at which we can see it after one firing has not been reached.  This continued devitrification at low temperatures can become great enough to be visible at the end of one or multiple slumps.

Credit: Bullseye Glass Co.

What can we do?

Clean all the glass before every firing very well.

·         Avoid mineralised water.

·         Final clean with isopropyl alcohol.

·         Polish dry at each stage with white absorbent paper.

 

Soak longer at lower temperatures.

·         Use longer soaks to achieve the slump.

·         Keep the temperature low.

·         Observe the progress of the firing with quick peeks.

 

Use slower ramp rates.

·         Slower rates enable the heat to permeate the glass.

·         Enables a lower slump temperature.

 

If there is any hint of devitrification after the first firing,

·      use a devitrification spray, or

·      provide a new surface.

• o   remove the surface by abrasion on sandblasting,
• o   cap with clear, or
• o   cover whole surface with a thin layer of clear powder.

·      Fire to contour fuse to give a new smooth surface.

·      Clean very well and proceed to slump.

Slower Ramps on Additional Firings

"Every time you fire a previously fired piece you need to slow down."

This is not accurate. If you have not changed anything significant, the annealing does not need to be extended.  The clearest example is a fire polish.  Nothing has been added. The physics and chemistry of the piece have not changed.  If only confetti or a thin frit/powder layer is added, nothing significant for scheduling has been added.  As nothing significant has changes the annealing used in the previous firing can be used again.

Of course, you do need to slow the ramp up rates on the second firing.  This is because you are firing a single thicker piece.  On the first firing, the pieces are individual and can withstand slightly faster rates. But on third and subsequent firings, if nothing significant has been changed, there is no need to slow rates further.

There is a post which describes this further.

"When adding more thickness more time is needed."

This is the occasion when the annealing soak needs to be extended.  Placing a full sheet of clear glass on the bottom, or less usually, the top, and taken to a full fuse, requires slower ramp rates.   The annealing time for a full fuse can be taken directly from the annealing tables for thick slabs.  

The fusing profile for any additional items has a strong affect on the length of the annealing soak.  If the glass is now of uneven thicknesses, and greater care in assigning ramp rates is needed.  The profile for the piece also determines the amount of additional annealing time required.  A sharp tack of a single additional layer will require annealing as for 2.5 times the total height of the piece at the start or the firing. A rounded tack will need two times and a contour fuse will require 1.5 times.  A full fuse can be carried out for the new total height of the piece without any multiplying factors.

 If the intention with multiple firings is to achieve a variety of profiles within one piece, a slightly different approach is required.  A blog post here describes the process.

The general approach to multiple firings is that unless there are changes to the thickness or profile of the glass, no changes in annealing time is required.  

 

Slow Rates to Annealing

"I have seen recommendations for slower than ASAP rates from the top temperature, but most schedules say 9999 or ASAP.  Which is right?"

Slow drops in temperature from top to annealing temperatures risk devitrification. Accepted advice is to go ASAP to annealing temperature to avoid devitrification forming.

Breaks do not occur because of a too rapid drop from top temperature to annealing. The glass is too plastic until the strain point has been passed to be brittle enough to break. On the way down that will be below an air temperature of 500˚C/933˚F.

credit: ww.protolabs.com

Different kilns cool from top temperature at different rates. Ceramic kilns are designed to cool more slowly and may need assistance to cool quickly.  This is usually by opening vents or even the door or lid a little. Glass kilns are designed to lose temperature relatively quickly from high temperatures. They do not need a crash cooling as ceramic kilns may need in certain circumstances.  Of course, crash cooling may be necessary for some free drops and drapes.

The length of the soak at annealing is determined by the effective thickness of the piece.  Tack fusing needs to be annealed for thickness as a factor of 1.5 to 2.5, depending on profile.

The extent to which you control the cooling to room temperature after the anneal soak is dependent on the calculated thickness of the piece you are cooling. The objective is to keep the internal temperature differential to 5˚C/10˚F or less to avoid expansion/ contraction differences that are great enough to break the piece. Those rates are directly related to the required length of the anneal soak.  Those rates can be taken from the Bullseye chart for Annealing Thick Slabs. The Fahrenheit version is is available too.

An example.  If you have a 2 layer base with 3 layers (=15mm) stacked on top for a rounded tack fuse, you need to fire as for at least 30mm. This will require controlled cooling all the way to room temperature.

• ·        The rate to 427˚C /800˚F will be19˚C /34˚F
• ·        The rate to 370˚C /700˚F will be 36˚C /65˚F
• ·        The final rate 120˚C /216˚F to room temperature.

You may need to wait a day before any coldworking. An example from my experience shows the necessity.  I checked a piece for stress a few hours after removing the piece from the kiln when it felt cool to the touch. It puzzled me that stress showed, although it didn't on similar pieces.  The next morning, I went to check if I misunderstood the reading. Now, a full 15 hours after coming out of the kiln, there was no stress.  The example shows that the glass internally is hotter than we think. And certainly, hotter than the air temperature.

In the temperature regions above the strain point, the glass needs to be cooled quickly. In the annealing region and below the glass needs to be cooled slowly.

More information is available in the eBook Low temperature Kilnforming.  This is available from Bullseye or Etsy

Heat Up vs Annealing

I am amazed by the effort put into ramp up rates, bubble squeezes, and top temperatures in comparison to annealing.  The emphasis on social media groups seems to be to get the right ramp rates for tack fuses and slumps, bubble squeezes, etc.  Most of the attention is on the way up to processing temperature.

The treatment of annealing and cooling is almost cavalier by comparison.  The attention seems to be on what temperature, and how long a soak is needed.  Then some arbitrary rate is used to cool to 370ºC/700ºF.

Annealing, in comparison to firing to top temperature, is both more complex and more vital to getting sound, lasting projects completed.  Skimping on annealing is an unsound practice leading to a lot of post-firing difficulties.

Annealing is more than a temperature and a time.  It is also the cooling to avoid inducing temporary stress. That stress during cooling can be large enough to break the glass.  This temporary stress is due to expansion differentials within the glass.

People often cite the saving of electricity as the reason for turning off at 370ºC/700ºF.  My response is that if the kiln is cooling off slower than the rate set, there will be no electricity used.  No electricity demands.  No controller intervention.  No relay operation.

Annealing at the lower end of the range with a three-stage cooling provides good results.  The results of Bullseye research on annealing are shown in their chart for annealing thick items.  It applies to glass 6mm and much larger.  It results from a recommendation to anneal at the lower end of the annealing range to get good anneals.  Other industrial research shows annealing in the lower end gives denser glass, and by implication, more robust glass.  Wissmach have accepted the results of Bullseye research and now recommend 482ºC/900ºF as the annealing temperature for their W96.  The annealing point of course remains at 516ºC/960ºF.

Bullseye research goes on to show that a progressive cooling gives the best results.  They recommend a three-stage cooling process.  The first is for the initial 55ºC/º100F below the annealing temperature, a second 55ºC/100ºF cooling and a final cooling to room temperature.

It is a good practice to schedule all three cooling rates.  It may be considered unnecessary because your kiln cools slower than the chart indicates.  Well, that is fine until you get into tack and contour fusing.  Then you will need the three-stage cooling process as you will be annealing for thicknesses up to 2.5 times actual height.

 

Of course, you can find out all the reasons for careful annealing in my book "Annealing; concepts, principles, practice" Available from Bullseye at

https://classes.bullseyeglass.com/ebooks/ebook-annealing-concepts-principles-practice.html

Or on Etsy in the VerrierStudio shop

https://www.etsy.com/uk/listing/1290856355/annealing-concepts-principles-practice?click_key=d86e32604406a8450fd73c6aabb4af58385cd9bc%3A1290856355&click_sum=9a81876e&ref=shop_home_active_4