Instructions: IQ’s Sapphire™ Wafers for Cutting GC Capillary Columns September 30 2016

Cutting capillary and making gas-tight seals is critical to good GC practice and making seals on silica press-fit connectors requires almost perfect cuts (unless you are using Pres2fit™ connectors). Cleanly cut ends are also important for CE for minimizing distortions of the electric field at the column ends.


The first thing that the separation scientist needs to understand is that not all silica capillary is the same. There are three major producers of GC and CE capillary around the world in Australia, Germany and in the USA. While very similar in materials of production, the differences in production methods lead to considerable differences in residual stress, geometry and chemistry. Some GC capillary is so oval that it can’t form an annular seal in standard press fit type connectors: not even our flaw-tolerant, universal Pres2fit™ (IQ P/N 3-2002S).

Residual stress in some capillary is so high that it is impossible to produce a clean, flat and normal cut (normal to the capillary axis).  The stress is a result of drawing capillary too fast or with too much tension (too cold). Capillary draw is non-ideal at best – typically 19mm X 25mm fused silica tubes are fed into a 2000°C, argon shielded graphite tube furnace and simultaneously pulled out at higher speed to reduce the diameter but within seconds it must be cool enough to apply a 20% polyamic acid solution (in n-methy-2-pyrrolidone) without boiling the solvent, drive off the solvent and dehydrate the polyamic acid to polyimide before the strand can contact the constant velocity capstan. (We actually apply 2 coats of polyimide before making contact with anything.)

Keeping everything centered in the draw line (melt furnace, laser micrometer, coating dies, curing furnaces) is easier when there is tension on the draw line but tension is directly proportional to residual stress in the glass strand. If there is not enough tension, the coatings will not be concentric and the capillary will be brittle. If there is too much tension the stresses “frozen in” the strand will favor axial fractures upon cutting.


A.A. Griffith developed the first theoretical model for fractures during WWI, inaugurating the science of fracture mechanics. As luck would have it, Griffith studied glass fibers. Telecom fibers are made of fused silica just like GC capillary and more telecom fibers were cut just installing Korea’s internet than all the GC capillary cuts made in history. Telecom cutting techniques are based upon fracture mechanics where almost every GC or CE capillary cutting guide is based upon no science whatsoever.

 There are a few facts about capillary cutting that need to be stressed:

 Capillary cuts are controlled fractures emanating from an induced defect: the score. You cannot “cleave” an amorphous material since there are no lattice planes.

  • The smaller the defect, the more force that is needed to initiate the fracture, but the flatter that fracture will be across the capillary face.
  • Multiple defects invite multiple fracture planes, the extreme case for which would be appropriate to call “shattering”.
  • Mode 1 cracks (opening mode like capillary cuts) propagate normal to the applied stress.


                        This isn’t coming from our lawyers, but from this scarred old chemist: wear safety glasses and disposable gloves while doing this. Sure, the bits of glass involved will rarely cut you, but a fragment in the eye really, really hurts. And the gloves (or finger cots) help you grip to the capillary securely, particularly where one side of the cut is a short section of material.



Figure 1: Proper orientation of ceramic wafers

The surface where you intend to work must be dust-free and hard. Clean the capillary with an alcohol wetted lab wipe for a few centimeter on both sides of the intended cut.

The goal is to produce a tiny (submicron) flaw to the glass wall of the capillary where the flaw is orthogonal to the capillary axis. The problem is that there are a couple of dozen microns of polyimide between you and your target and if the cut is intended to seal in a press fit connector, this polyimide must remain as undisturbed as possible.

Figure 1 illustrates proper orientation of IQ’s Sapphire™ wafer versus standard ceramic wafers. Sharper tools disturb less polyimide and induce smaller flaws than standard scoring wafers. IQ recommends the Sapphire™ diamond-honed scoring wafer (our P/N: 5-1001S1H) or a 30° diamond blade such as those sold by Element Six (The Netherlands). While holding the capillary firmly to the cleaned, hard surface with one hand, induce a flaw with the other, preferably with only a downward pressure while maintaining the sharp Sapphire™ or diamond edge normal to the capillary axis,

just nicking the glass surface. One may also draw the edge across the capillary

but, while easier, such motion does risk producing multiple flaws oriented in

slightly different planes and far more glass particles are generated.

                        Pick up the capillary and pull the ‘halves’ apart along the axis of the capillary (stress applied normal to the desired fracture plane). If the capillary refuses to break, put a drop of water on the defect and try again. If the capillary still won’t separate it is likely that the blade did not fully penetrate the polyimide or the flaw is too small. Note that all scoring tools wear with use and that sharpened blades are easily chipped: attempting to induce a precise flaw with a chipped area of a blade is fruitless. Drawing a chipped blade across a capillary will usually yield a poor result.


      Figure 2: Cracked capillary in press fit union    

                          The axial stress that is required to propagate a fracture is inversely proportional to the flaw size and directly proportional to the flatness of the resulting fracture face. Examine the cut end of the capillary with a magnifier before use to insure the cut is adequate to the need and to insure no axial fractures occurred (residual stress induced). Figure 2 is an image of a stress induced crack in GC capillary that was installed into a press fit connector (not ours) with polyimide resin (not recommended).

Look for loose glass particles that could enter the capillary opening and potentially damage the column; flicking the cut end of the capillary a few times may serve to dislodge any larger particles that may be electrostatically adhered to the glass or polyimide.  Glass shards with diameters on the order of the thickness of stationary phase, that get into the opening of the capillary, may be driven along the length of the column by carrier gas or buffer, impacting the capillary inside wall along the way and leaving just the right size flaws to make the column brittle and useless. Even where the possibility is remote, precautions against such damage should become routine.


Wipe the capillary again with an alcohol wetted lab wipe.

Figure 3: Glass shards from cutting


Press fit type fittings are rarely reused, but if you are reusing any type of connector, you must insure that it is also free of potentially damaging particles before installing the column, particularly if it held a fractured capillary at any time in the past.

For press fit connectors, merely insert the capillary until it stops on the glass cone wall, then apply additional force to “seat” it securely; the more force used, the greater the area of the seal that will form but the greater the risk of amplifying residual stresses and cracking the capillary. Such friction fit connections will hold 5000psi, although such extremes are definitely not recommended. A brown band that appears wide than the polyimide on the capillary will appear where the polyimide is in intimate contact with the conical glass bore: this is the seal area. If the band does not extend 360⁰ around the capillary, the capillary is too oval and a seal has not formed: the connection will leak and no amount of polyimide resin will fix it.

It is not necessary to use polyimide to seal a capillary into a press fit fitting, much as “Teflon tape” is useless on Swagelok® fittings: the seal is formed elsewhere. Polyimide resin is only useful where press fit installation is such that significant vibration or other motions could potentially

dislodge the capillary from the connector: it will not seal a leak. The ‘polyimide’ that you use should be the same resin as the polyamic acid that used to make the capillary (IQ P/N PI2525-5G). Semiconductor grade polyamic acid is expensive, in part because it is filtered to remove sub-micrometer sized particles that would interfere with semiconductor processing. Other grades, such as those used to coat transformer wires, may have the same or similar polyamic acid structure and chemistry (although typically lower in average molecular weight), but are not filtered and may harbor damaging particles. These lower grades of polyamic acid resin typically use a mixed solvent system to lower costs even further, with additives identified as aromatic hydrocarbons, xylene and other common organic solvents. Check the MSDS. If it lists any solvent other than NMP, it is probably not semiconductor grade or it’s not BTDA-ODA-MPD: it’s definitely not the super-secret formulation used to make capillary (HD Microsystems PI2525).


Figure 4: BTDA-ODA-MPD copolymer formation followed by imidization