|
At the operating temperature of the lamp sodium is chemically very active. Ordinary glasses (even quartz) are rapidly stained brown which blocks out light. The stained areas also have a different coefficient of expansion from the unstained material, and differences in expansion may then cause the glass to crack. In 1920, a special sodium resistant glass - borate glass - was developed by A.H. Compton. However, pure borate glass is unsatisfactory in lamp manufacture. It is known as a ‘short glass’, meaning that it has a short working temperature range. On heating, the glass changes from the rigid condition to fluidity very quickly, thus making it rather difficult to process. The earliest sodium lamps could only be manufactured with the assistance of the most skilled master glassblowers who were adept at handling this glass.
The problem was overcome by using a 2-ply glass tube. A very thin layer (about 0.02mm) of barium aluminoborate glass is blown onto the inside of an ordinary soda-lime glass tube. The soda glass can be worked very easily on automatic equipment, and it acts as a support for the borate lining which is so thin that it is almost insignificant. This allows the sodium-resistant ply tubing to be formed easily using standard glassworking methods. Figure 27 shows a cross-section through some Ply tubing.
Figure 27 - Cross section of ply tubing employed in low pressure sodium lamps
The coefficients of thermal expansion of the borate and soda-lime glasses are not perfectly matched, and this can cause problems of cracking through the introduction of permanent thermal stresses. It is necessary to provide longer annealing schedules for ply tubing than for other glass types. In addition, it is important to maintain uniformity of the borate coating thickness around the tube circumference, so that radial stress differences are minimised. The most critical operation is bending the tube. During this process the glass wall thickness naturally decreases on the outside of the bend, and increases on the inside. This effect must be minimised by correct location of the melt in the U-shaped mould and by the glass being stretched slightly before moulding.
If there is any unevenness in the thickness of the borate layer, sodium corrosion will be more of a problem in the thin areas. This will cause early lamp failure by excessive light absorption arising from stained glass, or glass cracking. Even if the glass is made well, good lampmaking practice is also required because the borate coating can be easily damaged. This glass is highly sensitive to moisture and is readily attacked by it. For this reason, borate glass is generally delivered to a lamp factory in heated trucks, and stored in a heated room to prevent moisture condensing in the glass. The temperature must also be kept constant, since it can otherwise result in expansion and contraction of the plastic packaging material which may draw moist air into the tubes.
Another major drawback of borate glasses is that they tend to clean up argon, a gas which is essential for easy lamp starting and long life. Many decades of development have now produced a glass which has good sodium resistance and a relatively low argon cleanup rate. Although the glass is not yet perfect, it has been improved to the state where neither of these two failure modes cause significant problems. The adhesion of liquid sodium to borate glasses has also been thoroughly investigated, so as to provide a glass type in which the molten sodium is less likely to flow inside the tube if it is not kept absolutely horizontal. In early SO lamps, if the discharge tube was not perfectly level then molten sodium could flow down to one end of the lamp where it would form large light-blocking mirrors, and lead to sodium depletion and reduced efficacy at the other end of the lamp. | 
