Deuterium lamps produce intense ultraviolet (UV) radiation at wavelengths between 190 and 400 nm via plasma discharge in an atmosphere of deuterium. Light with a wavelength below 190 nm is not readily transmitted, as it is absorbed by the lamp’s quartz envelope.
These lamps, as is the case for any light bulb, have a finite lifespan. The operating life can be defined as the total number of hours for which the lamp can be used for meaningful work. This useful lifetime for a trace-level analysis is likely to be shorter than the lifetime for routine analysis because of the lower level of quantitation required.
What affects lamp life?
However carefully it is used, the lamp will always be subject to normal degradation. Its metal coatings will evaporate, and there is a filament coating reaction with the quartz envelope. Furthermore, solarization of the transparent sheath material occurs, with shadowing from intense radiation.
Some degradation is related to the way the lamp is used. The number of times the lamp is ignited increases stress on the lamp filament and is inversely proportional to its operating life. However, leaving it on continuously will tend to decrease its useful life by about threefold, assuming the instrument is used eight hours a day. It is advisable to shut the detector lamp off when not being used for extended periods of time.
Touching the lamp with bare fingers can reduce its lifespan by creating hot spots on the outer envelope from oily residues. Physical shock also has an effect, whether it is caused by moving or jarring the detector while the lamp is on or powering down the detector and not allowing the hot lamp to cool down before it is turned on again.
Lamps also have a limited shelf-life before installation, so it is important to check the date of manufacture before installing the instrument. Additionally, it is important to condition the incoming voltage to an instrument to prevent spikes or swings that may also damage the lamp.
Indicators of lamp problems
An experienced analyst will recognize the signs that a lamp might not be performing perfectly, including baselines that feature irregular, high, or periodic noise. There are several reasons why this may be the case, but, if other factors—including air bubbles in the system, temperature swings or pump pulses, a contaminated flow cell, or if immiscible solvents are being used—then an aging lamp may be the cause.
Another indicator is that the peak response is lower than expected. Again, the flow cell might have been contaminated, or the injection volume might be insufficient; but it may also be indicative of the lamp’s diminishing performance.
Characteristics of the lamp itself can also have an impact. The lower the wavelength, the more degradation in output can be expected. This may mean that an aging lamp that works almost perfectly at 340 nm and is acceptable at 280 nm could be next to useless at 210 nm.
Of course, if the signal from the analyte is sufficiently large, then, in reality, the substandard energy output from the lamp may not be a problem. Lamp energy becomes important if the signal is small. If the signal-to-noise ratio is low, then the greater impact of the baseline noise may alter the detectability of trace analytes due to the compromised output from the lamp.
When to replace a lamp
Clearly, if the lamp does not illuminate when the power is switched on, it must be replaced. It may still be emitting UV light, but the analytical results will be obviously atypical. But it is rarely that clear-cut, and a strategy to determine its replacement is important.
If the peaks of interest have a strong signal, in other words, in excess of 100 mAU/mV, then it is reasonable to wait until the lamp fails to ignite until it is replaced. This will typically be 4,000–8,000 hours. However, if trace analyses or impurity checks are being performed, a procedure should be in place to evaluate the signal-to-noise ratio of the instrument at the set wavelength. If this is suboptimal, then it is likely that a new lamp will need to be purchased as a consumable if its warranty has expired, and the unit would then need to be recalibrated and qualified.
Another option is to monitor the lamp’s performance on a regular basis. This approach may provide some indication of an impending lamp failure if the checks are sufficiently frequent. The problem with this strategy, however, is that once the lamp starts to fail, its output will decrease exponentially and the failure will occur extremely rapidly.
Some instrument systems include an early maintenance feedback feature that tracks the hours of use. For an analyst, it is probably not feasible to rely solely upon this, given the unpredictable nature of the operational life of the lamps.
Alternatively, an intensity test can be performed to monitor the status of the detector lamp, although this is not always conclusive without further confirmation testing. If it does fail this intensity test, the first step is to rule out all other common causes, such as solvents, bubbles, dirty flow cells, or optical issues.
To differentiate detector problems from flow cell issues, and to perform a detector cell test, a test cartridge would need to be purchased, which can be expensive. If the lamp is pinpointed as the cause of the intensity test failure and is replaced, then it will need to be recalibrated and qualified.
Another approach, although more expensive, is to replace the lamps at regular intervals, even if the equipment is working properly. For example, UV lamps could be replaced after a certain number of hours in use—say, 4,000 hours. The lamp hours would need to be monitored, and there would still be no guarantee that failures would not occur before the lamp is replaced.
Alternatively, the lamps could be replaced on a regular cycle, for example, every three months, which represents about 2,000 hours of continuous use. Again, there is no guarantee that lamp failures would not occur before the new lamp is installed, but it would be under warranty.
The material cost of exchanging lamps on a quarterly cycle is substantial. Carrying out three additional changes a year, on top of the periodic maintenance replacement for 25 instruments with long-life deuterium lamps, would be close to $100,000 per year. In addition to the consumable costs, each change would require the detectors to be recalibrated and qualified, adding greatly to the cost incurred in both dollars and time.
Selecting the right strategy
When considering the lamp replacement strategy, the first question to ask is: How many HPLC system failures can truly be attributed to the failure of a deuterium lamp in any given year? Additionally, can these lamp failures be substantiated through vendor service reports or logbook entries? Do guidelines exist for shutting lamps off when not in use, and which has the greater impact on lamp life—ignition frequency or continual operation?
The decision will also depend on whether the instrument is typically used for trace and impurity analyses, where background noise has a greater impact on accuracy. Are signal-to-noise, limit of detection, and limit of quantitation determinations performed, monitored, and documented? Are detector optimization parameters such as bandwidth, slit width, reference settings, and response times reviewed or changed when new methods are brought on line?
Ultimately, the strategy will depend on the evidence and the risk-to-reward ratio. A large company may be less concerned about the cost of replacing a $600 lamp every quarter, but a smaller one may choose to take the risk of replacing on a less frequent basis.
Robert Pritchard is analytical instrumentation specialist, and Mark Shapiro is director, analytical research & development, Cambrex, 4180 Mendenhall Oaks Pkwy., High Point, NC 27265, U.S.A.; tel.: 336-841-5250; e-mail: ; http://www.cambrex.com