Sciencemadness Discussion Board

Sealing wires inside glass

Antwain - 9-11-2007 at 03:26

I am looking for adivce on how to sucessfully seal wires inside glass, such as glass tubing. I know that many people here have done this.

I have had 3/3 failures on this front. They are all leaking. And they all suffered oxidation of the metal during the sealing process.

Is is only possible to seal a single wire of circular crossection inside glass? two of the failures were with 2 strands of platinum wire, and the oxidation was of the copper wire it was connected to. The other failure was with 8 strands of titanium wire (a lot, I know) which was only superficially oxidised, but the copper wire several inches up the tube was substantially oxidised.

Both titanium and platinum have very similar coefficients of expansion, and they are nearly the same as glass so that should not be the problem.

Any advice appreciated.

YT2095 - 9-11-2007 at 04:25

I rough up the wire a little 1`st (fine grade sand paper) then you have 2 ways that I found that works well.

1. gravity melt insert the wire while still open and then Suck.

2. part melt, insert the wire and then crimp with hot tongs.

YT2095 - 9-11-2007 at 04:28

also, it`s best to practice with Copper or something 1`st, I certainly would not think about using my Good metal until I had made a few perfectly with ordinary wire.

Antwain - 9-11-2007 at 05:37

Ahhh, so it wont just magically just 'seal itself up' then. I kind of just let it flow in while twirling it a bit. For some crazy reason I just figured that the glass would have enough affinity for the metal that it would 'wet' it like water wets glass through capillary action. Not so, apparently.

YT2095 - 9-11-2007 at 05:41

no, not really.

it will need "forcing" a little, either with Suction or a clamping while hot, and it`s a good idea to rough up the wire a little so you get a good surface area for purchase.
remember to keep your clamps/tongs just as hot as the glass though, and a little bend whilst its on helps quite a bit too. then just get it all hot and annealed again.

Maya - 9-11-2007 at 09:15

<<< For some crazy reason I just figured that the glass would have enough affinity for the metal that it would 'wet' it like water wets glass through capillary action. Not so, apparently>>>

The only metal that will 'WET' glass is indium

Mr. Wizard - 9-11-2007 at 09:54

Take into consideration the difference of expansion of the metal and the glass. A large cross section of metal is going to expand a lot more than a thin layer of glass, leading to a crack and a leak. Try to flatten the wire where it will be surrounded by glass. Another idea is to coat the thin metal wire with a glass bead before sealing it in a glass tube. Clean the thin copper wire with fine sand paper and fuse a little sodium borate on the wire. It acts as a flux and keeps the metal clean. Once you have a nice clean wire glass bead, you are ready to put it into the tube. You might even try using a mixture of Borax (sodium borate) with powdered glass to get a stronger bead. Another trick would be to use an old light bulb. The sealed in wire can be fused to your work as a unit. Old Christmas tree bulbs, especially the tiny colored ones are almost made for this work.

indigofuzzy - 9-11-2007 at 18:16


The only metal that will 'WET' glass is indium

Or Gallium.

RxnJackson - 9-11-2007 at 18:26

depending on the wire material, is there acid or something that may prepare the wire for a better bond to the glass (Ie flux with sautering), or is the glass just too inert, purely a mechanical bond?

12AX7 - 9-11-2007 at 21:09

I can't imagine you could possibly put anything else in there that would help. Do use a reducing flame around copper.

Possibly, a lower-melting, more reactive (i.e. fluxing) glass -- like fused borax -- could be used, but it probably wouldn't match the expansion of the soda-lime (or other) glass around it, plus it may dissolve in solution.


P.S. Aluminum will also wet glass. Which really, I should say they'll wet each other at that temperature.

Antwain - 10-11-2007 at 03:01

Thanks for the tips guys.

Wets is a relative term. It can mean many different things, such as that the interaction between between x and y is greater than the interaction between x and x or, y and y. Or it can refer to the angle that the meniscus makes between a liquid and solid. Or it can refer to relative affinity, such is will something bond more strongly to glass than say air.

Like I said, it was a silly assumption, but you do these things and learn for yourself sometimes. (or if you are smart you read first :D )

garage chemist - 11-11-2007 at 19:06

Platinum wire is supposed to be especially easy to seal into glass, it was used in the first lightbulbs for the airtight seal of the electric connections (nowadays, a special alloyed steel is used whose thermal coefficient of expansion matches that of platinum and therefore that of the glass).
Its really mostly about the thermal coefficient of expansion, and platinum happens to have nearly the same one as soda-lime glass.
The second criterium is resistance against oxidation when heated in air, and nothing is better than platinum in this regard.

I have sealed platinum wire into glass two times now, both times a success.
I suggest that you continue heating the glass for some time after it has molten around the wire in order to make sure the glass bonds to the platinum and wets it.
Copper or other metals should not touch the glass! Leave about 2cm of the Pt wire between the seal and the connection to the copper wire.
Let cool very slowly, ideally in an annealing oven to eliminate stress, or cover the glass seal with soot from a smoky propane flame while it is still glowing red hot and then pack it into rock wool for slow cooling.

This is just a tip, I didnt have to do this with my platinum wire elctrodes, they were watertight after the first try.

I used a glass tube with one end drawn into a tip, like a pipette.
The thinner glass at the tip makes sealing to the wire much easier since the amount of glass that has to be softened is much smaller, and the tip already has an inside diameter close to the diameter of the wire.
This "pipette trick" worked perfectly for me.

[Edited on 12-11-2007 by garage chemist]

woelen - 11-11-2007 at 23:31

For me, the making of the electrodes was very simple. I first soldered the platinum to the copper wire, and then I took a glass tube with an inner diameter of 5 mm or so. I put the wire in it and heated the tip of the tube in a small hot flame, while slowly urning around the glass tube. Although the droplet of glass at the end of the tube is not perfectly round, it is sealed very well. I have used these electrodes quite a lot already, and they still did not corrode inside the tube.

Below follow pictures of the two electrodes I have. One has somewhat thicker wire than the other.

pt_anode1.jpg - 25kB

pt_anode2.jpg - 26kB

Edit(woelen): Made links to pictures work again.

[Edited on 30-7-16 by woelen]

Twospoons - 12-11-2007 at 12:21

You might have better luck using the kovar wire from inside an ordinary household lightbulb. This has the advantages of being cheap, readily available, and thermally matched to soda glass. If you are using pyrex you will need a different kovar alloy (there are several, with different expansion coeff's)
For electronic glass-metal seals, the wire was often copper plated over the seal region. The soft copper layer allows a small amount of shear without breaking. IIRC ordinary steel wire can be used like this.

12AX7 - 12-11-2007 at 14:04

It's my understanding that copper-plated kovar is easier to seal.

For sure, all my vacuum tubes, where the glass seals are visible, are sealed at shiny pink copper, but copper through and through they ain't.


dann2 - 13-11-2007 at 19:02


The following attached file is OCR'ed from 'Creative glass blowing'. A scientific american book.


Attachment: Metal to glass.html (16kB)
This file has been downloaded 2862 times

Antwain - 13-11-2007 at 20:55

@dann2 - there is something horribly wrong with that link. It just brings up garbage.

Well I think the fundamental problem that I had more than anything else was not leaving enough length of wire. I am about to spend a stupidly large amount of money to get maybe a meter of descent wire at a good price and then I will have enough to do the sealing properly. The problem was that the heat was going up the wire and wrecking havoc on the copper and solder, so I couldn't heat it for as long as I wanted.

Nixie - 14-11-2007 at 13:49

Are the best glass seals directly metal to glass, or can some improvement be had from putting some type of cement or something in between? I've been thinking DIY vacuum tube as I may have temporary access to a high vacuum system later this year.

dann2 - 14-11-2007 at 13:53


When you say the link brings up only garbage are you referring to the content???:D
The file works OK for me, anyone else with a problem with this file. It is simply a common or garden HTML type file.

I copied and pasted below.

Glass-to-metal seals

Frequent occasion arises for sealing glass to metal. For example, a handy dissection needle can be quickly improvised by pushing the eye-end of a heavy darning needle into the molten end of a glass rod. Holders of various kinds can be similarly made of glass rod and short lengths of wire or slivers of sheet metal. Almost any kind of glass and most of the base metals, if not more than a millimeter in diameter, can be so joined. The bonds have reasonable strength. In general, however, the thermal coefficient of expansion of metal greatly exceeds that of glass. Copper, for example, expands 167 parts in 10 million for each degree Centigrade of temperature rise, whereas soft glass, such as Corning code 0080, expands 92 parts in 10 million for each degree, and Pryex 7740 expands only 32.5 parts. As the joint cools, the metal contracts more than the glass and tends either to pull away from the bond or to set up a severe strain in the glass. A steel rod 1 millimeter in diameter, if sealed to soft glass, would contract about 5 microns upon cooling from the annealing temperature of soft glass to room temperature. Although the seal would doubtless appear mechanically strong, the chances are great that the metal would pull away from the glass, a matter of no consequence in the case of a dissection needle. But a seal even slightly cracked could not be tolerated in an incandescent lamp bulb or other application involving high vacuum. The joint would leak. For sealing leads or other metal parts into vessels that must be exhausted and maintained at low pressure, the thermal coefficients of expansion of the metal and glass must match within about 1 part per million. Moreover, when the seal is made, the molten glass must "wet" the metal. This is an easy requirement. The metal need only be oxidized slightly at its point of contact with the glass. The glass dissolves the oxide and in effect unites with the metal in a smooth chemical transition. The two materials are compatible because glass is composed largely of metallic oxides.
The thermal coefficients of expansion of only two metals approximately match those of common glasses. Platinum, which expands 90 parts per 10 million per degree Centigrade, is an excellent match with the soft glasses such as soda lime and lead that expand from about 87 to 92 parts per 10 million. Tungsten expands 48 parts per 10 million per degree Centigrade, compared with Corning 7740 Pyrex at 32.5 parts and 7720 Pyrex (better known as "Nonex") at 36 parts per 10 million per degree Centigrade. Uranium glass (Corning No. 3320) expands 40 parts per 10 million per degree Centigrade. As suggested by these data, platinum of any size can be sealed directly to the soft glasses to form a vacuum-tight joint. A similar seal can be made between tungsten and either 7720 glass or 3320 glass. These glasses, in turn, seal to 774'0 Py rex. Oddly enough, copper, which expands 168 parts per 10 million parts per degree Centigrade, can be sealed to any glass. This is made possible by the great ductility of copper. Plastic flow in the metal relieves the stress of the cooling seal before it reaches the breaking point of the glass. The technique of making such seals was developed in the early 1920's by William G. Housekeeper of the Western Electric Company.
Subsequently, a number of alloys with thermal coefficients of expansion that closely approximate those of specially compounded glasses, as well as those of conventional soft glasses, have been compounded. Most, such as Kovar, developed by the Westing-house Electric Company, and Fernico, a product of the General Electric Company, are compounded principally of nickel, cobalt, and iron. They made possible the development of the "all-metal" electron tubes. Another is Sealmet, a product of the Higrade Sylvania Corporation. Unfortunately, these alloys require extended heat treatment at about 1000 degrees Centigrade (1832°F) in an atmosphere of hydrogen bubbled through water, and they must be used within a few hours of this degassing procedure. If the alloys are not so treated, fine bubbles appear in the seal that weaken the bond and cause leaks. Because few beginners, students, or small laboratories have access to thermostatically controlled hydrogen furnaces, we will not discuss the use of these alloys. Those who wish to pursue the topic are referred to the excellent professional reference, Scientific and Industrial Glass Blowing and Laboratory Techniques, by W. E. Barr and Victor J. Anhorn (Instruments Publishing Company, Pittsburgh, Pa. 1959;.

Dumet seals

One alloy that is compatible with the soft glasses requires no degassing. This is a copper-coated composition of nickel and iron known as Dumet. The alloy is used for sealing leads into electric lamp bulbs, in electrodes of neon signs, and so on. Dumet comes in the form of wire in sizes ranging from about 20 to 28 gauge, precut to a length of about 6 centimeters. Normally, the wire is inserted through the end of a glass tube of appropriate diameter and sealed by softening the glass and squeezing the ends of the glass and the wire together. The result is known as a "press seal."
To make a seal using Dumet leads, first soften the end of a tube as in Figure 6-13, a. Crimp but do not comletely close the softened end (6-13, b). Insert the Dumet leads. (It should be noted at this point that some device will usually be connected to the inner ends of the leads—perhaps a cylindrical electrode, a filament, or a miniature metallic crucible containing a substance such as an alkali metal. If so, the metal assembly may be inserted in the open end of the tube with the Dumet leads facing the partially closed end.) Then incline the tube so the assembly slides to the partially closed end with the leads protruding (6-13, c). Soften the glass, squeeze into intimate contact with the metal, stretch about 3 millimeters, and then heat strongly to assure that it wets the metal (6-13, d). Surface tension will thicken the molten glass (6-13, e). Restore the seal to its former thickness by squeezing the end with a pair of tweezers (6-13, /). Anneal (6-13, g).

Sealing tungsten to borosilicate glass

Gas-tight seals between tungsten and borosilicate glass are not quite as easy to make as those between Dumet and the soft glasses. The metal itself tends to be leaky. Tungsten is reclaimed from its ore as a powder that is subsequently converted into billets by sintering. The billets tend to be slightly porous. Wire drawn from them may contain microscopic channels capable of conducting gases. To make the wire gas-tight another metal such as nickel may be butt-welded to the ends. Tungsten wire used for conducting electric current through glass is usually made by welding a flexible copper lead to one end and a nickel wire to the other (Fig. 6-13, h). Such leads are available commercially.
Tungsten oxide does not enter into solution with molten glass as readily as does copper oxide. Yet, the glass must dissolve the oxide down to the metal if the seal is to be gas-tight. A thick, partially dissolved film of oxide may leak. In addition, tungsten must be degassed before it is sealed to glass, or bubbles will form at the interface between the glass and the metal. The bubbles weaken the seal and may encourage a leak. Finally, the thermal coefficient of expansion of tungsten differs so much from that of Pyrex glass that the two must be joined through a glass of intermediate expansion, a so-called "sealing" glass, if the metal is more than 0.5 millimeter thick.We recommend the use of commercially prepared tungsten leads, those that have been plugged by butt-welding. The tungsten portion of the lead is degassed by bringing the metal to white heat in an oxygen-enriched fire. After the metal cools to a dull red, rub the tungsten against a lump of potassium nitrate and wash thoroughly in distilled water. This treatment should remove the oxide. If blotches of dark oxide remain, repeat the treatment until they disappear. Then reoxidize by heating the tungsten to dull red for a few seconds. The thickness of the resulting oxide film varies with both the temperature and the heating time, quantities that must be determined by experiment. A film of correct thickness is easy to identify after the seal has cooled. The color of the interface between the metal and the glass of a good seal ranges from yellow to reddish-brown. A film of oxide that is too thick causes a black interface. Such seals may occasionally be corrected by maintaining the glass in the molten state for a minute or so. If the treatment is successful, the black interface will turn reddish-brown. Conversely, oxide films of insufficient thickness do not make good seals. The glass does not adequately wet the metal. The interface has the color of the unoxidized tungsten.
The metal must always be coated by a glass that has a thermal coefficient of expansion intermediate between that of tungsten and that of Pyrex. If the thickness of the wire does not greatly exceed 1 millimeter, we use Corning 3320 (uranium glass) or Corning 7720 (Nonex). For thicker wires, we fuse 3320 glass to the metal, 7720 glass to the 3320, and Pyrex 7740 to the 7720—a structure known as a graded seal. (See the table at the end of Section I, page 18.)
To apply the sealing glass to the wire, first draw an 8-millimeter tube of 3320 glass to an inside diameter slightly larger than that of the metal. Cut the small tube to the same length as the tungsten to be covered. Slip the tube over the lightly oxidized wire and fuse it in place by concentrating the heat on one end of the glass tubing. As the glass shrinks into contact with the metal, move the heat gradually to the other end (6-13, 3). All air must be squeezed from the interface by the shrinking glass. Cool the seal slowly to minimize the development of strain (6-13)
Tungsten so coated may be used in a press seal, just as Dumet is sealed to soft glass. Alternatively, a single lead may be sealed through the wall of a bulb or other apparatus by applying the sealing glass to the metal in the form of a bead. The bead is then sealed into a hole of smaller diameter in the Pyrex. Incidentally, tungsten is a relatively brittle substance at room temperature. It may be bent if heated to a bright red.

Housekeeper seals

Copper in any one of four shapes— wire, thin sheet, tubes, or disks—may be sealed to either the soft or the borosilicate glasses. If you are using wire, first flatten the section that will come in contact with the glass to a thickness of not more than 0.5 millimeter and then file the edges to knife-sharpness (Fig. 6-14, a, b, c). As viewed in cross section, the flattened and sharpened portion may take ths form of a parallelogram (6-14, d). Heat the piece until the color changes to a reddish-brown, indicating the formation of a light film of oxide, and immediately paint it with a concentrated solution of borox or drop it into the solution (6-14, e). For a solution, the Borateem variety used as a household detergent works nicely. When dry, the wire should be uniformly covered by a white film of borax. The metal may then be incorporated into a press seal. Confine the molten glass to the flattened portion of the wire and concentrate the fire more on the glass than on the metal (6-14, /). Heat converts the borax into a form of glass that not only helps to dissolve the oxide but shields the copper from excessive additional oxidation, the property that accounts for the usefulness of the substance as a welding flux. Sheet copper up to 0.5 millimeter in thickness and 25 millimeters in width may be similarly sealed.
Copper tubing is prepared for sealing to glass tubing by filing or otherwise sharpening the end of the copper to a feather edge at an angle of approximately 10 degrees (6-14, g). A holder of some sort must be improvised that will plug the tubing and enable you to manipulate the hot metal. We use a tapered plug of Transite into which a handle of 8-millimeter glass rod is cemented with sodium silicate. Coat the heated tube with borax and then seal into a glass tube just large enough to slip over the end of the metal. The glass must not extend beyond the point at which the thickness of the copper exceeds 0.5 millimeter (6-14, h). Then burn off the glass just beyond the end of the metal and blow off the resulting bulb (6-14, i). Coat the inner surface of the metal by spinning the unattached portion of the glass over the edge by by means of a flaring tool (6-14, /). Promptly seal the glass coating to any desired tubing before the metal cools (6-14, k). Pyrex can be joined successfully to soft glass via a copper sleeve, and successful seals as large as 15 centimeters in diameter have been made. They are rather weak mechanically, but vacuum-tight. When joining Pyrex to soft glass by this technique, make the Pyrex-to-metal seal first.
Occasion also arises for closing the end of glass tubing with a metal disk in some types of gas discharge tubes and for sealing electrical conductors of largo diameter into evacuated apparatus. Copper disks of any desired diameter and up to 0.5 millimeter in thickness seal readily to the flared ends of glass tubing. Just heat and lightly oxidize the disk, drop it while it is hot into the concentrated solution of borax, and dry (Fig. 6-15, a).
Place the disk on a hot block of carbon and bring the softened edge of the flare lightly into contact with the copper. The glass will melt the borax and adhere to the tacky film (6-15, b). Pick up the adhering disk with the glass and make a conventional butt seal to a second tube that has been flared to the same diameter (6-15, c, d). Pull a point in the second tube close to the disk (6-15, e). By blowing and simultaneously pulling, expand this point into a long, thin bulb (6-15, /). Strike off the bulb and remove the jagged edges by stroking with a piece of wire screening that has been tacked to a wooden paddle (6-15, g). Fire-polish the edges and anneal (6-15,).
A conductor can be supported by the disk, of course. Simply drill the disk, insert a conductor of the desired size, and braze it in place with an alloy that melts at a reasonably high temperature, such as silver solder. Drop the hot brazed assembly into a pickling solution consisting of one part of sulfuric acid in nine parts of tap water. The pickling solution removes the excessive oxide formed by the brazing operation. Rinse the piece thoroughly to remove the acid. Then reheat to oxidize lightly and coat with borax (6-15, i). Support the metal assembly in the flared glass tube by means of a roll of asbestos tape, as illustrated Make the seal and form the glass ring on the onlside of the disk as previously explained.
Ingeneral, Housekeeper seals are relatively weak, mechanically. They are also subject to attack by a number of chemicals as well as by Mercury. They may be electroplated easily, however, and thus protected against selected substances.



dann2 - 25-8-2008 at 13:06

Book for sale on ebay.

I have no connection with the sale.


ordenblitz - 25-8-2008 at 18:04

I had lots of problems making glass to wire seals with a propane torch. I switched to MAPP gas and suddenly it worked every time.