Multiplexing UV-NIR Spectroscopy Measurements

Dr. Gert Noll, TecUSA

Optical coating of large areas has become an increasingly important step in materials processing. Achieving the highest possible performance and respective yield from large panes of architectural glass, solar cells and thin-film layers requires permanent control of the optical performance during production. Several measurement spots across a web are needed to determine optical performance.

For high-throughput process control, only fast-readout spectrometer systems based on detector arrays are appropriate because only they can acquire a full spectrum in milliseconds or less (if a sufficient amount of light is detectable in that short period of time). Such detector array spectrometers have the further advantage of no moving parts, rendering them highly stable devices if assembled by someone with experience in mounting and cementing techniques. This is the key to generating data report after data report in no time with high reproducibility. Since there is no scanning grating or other wavelength-selective element that would have to be moved, all detector pixels see light at all times; no photons are filtered out and, therefore, wasted. This makes a detector array-based spectrometer a very efficient device, ideal for achieving short measurement times.

This spectrometer system has a four-channel fiber optic multiplexer. The prices for optical multiplexers vary with the number of channels required.

To measure multiple spots for transmission, reflectivity, color or thickness, a traverse typically is used, transporting the measurement head – or the full spectrometer system – from spot to spot, one after the other. However, the drawbacks of such a setup are the problem of moving larger masses in a reliable way, the time required to move the measurement head to the individual spots and the high cost of traverse setup, including installation.

The perfect solution is to install a complete spectrometer system for each measurement spot. With a modern computer and software, it is no big problem to control more than 10 USB or Ethernet interfaces at a time. But the drawback of the high cost remains, especially if high-performing spectrometers are required.

Multiplexing

To save money, various multiplexing solutions can be applied:

1. Electronic multiplexing

Electronic multiplexing is based on a special electronic device that acts as a kind of operating electronics input expander. One set of electronics consisting of an analog-to-digital converter and a computer interface can handle various spectrometers (Solutions for up to eight N-type metal oxide semiconductor-based spectrometers are available). While using a multitude of spectrometer modules, the saving is on the electronics. Another advantage of this solution is that all spectrometers can be read out nearly simultaneously with a fixed time relationship between all channels.

No optical parts are moved, and the amount of light detected is not compromised. All spectrometers are sensitive at all times and are constantly collecting incoming photons. The overall measurement time can be as short as it takes to read out all spectrometers, which can be in the range of 10 ms at eight channels of spectrometers, based on 256-element detectors.

2. Optical multiplexing with a blocking device

If a multitude of spectrometer modules is too expensive, optical multiplexing can offer a cost-effective alternative. The most reliable way to do this is to use a setup that does not require any moving optical part. Such a design can be based, for example, on optical fibers that diversify the illumination light into various paths. Mechanical shutters or a chopper wheel are used to block and open the individual paths. Lightguides transport the light to the individual measurement spots and pick up the light to be analyzed; the light eventually is forwarded to one spectrometer. One after another, the individual paths are unblocked, and the measured spectra are allocated to the related channels, thus allowing differentiation among the various measurement spots.

The reproducibility of such a setup is as good as that of one with individual spectrometers, since no optical part is moved. Drawbacks are slower data acquisition, because the mechanical blocking mechanism has to move, and loss in light, because each measurement spot is observed during only a fraction of the overall time required to take the data. The more channels have to be used, the more light is lost. While it is a feasible technology for two or three channels, losses may be too large for four or more. In this conjunction, it is important to recognize that losses in throughput must be covered by a longer measurement time or by sacrificing signal quality. While still sufficient to perform a transmission measurement, it might not be good enough for a reflectance measurement. The applicability depends strongly upon the task.

With a mechanical shutter approach, up to five measurements per second can be achieved. Switching times can be significantly lower if a chopper rotor is used to block the light, but a good synchronization between the chopper wheel and the data acquisition has to be guaranteed. This requires direct access to the control electronics. The limiting factor might be the amount of light available.

3. Optical multiplexing with a switching device

To overcome the introduction of high losses by a shuttered device, devices that redirect optical light can be used. One convenient solution is a fiber optic switch (or multiplexer). These switches move fibers – mostly driven by piezoelectric elements – so that, for example, the light from various input fibers coming from the sample heads is directed, one fiber at a time, to one output fiber leading to the single spectrometer.

Shown here is a six-channel spectrometer cassette with electronic multiplexer. Photos courtesy of Tec5USA.

Such switches are available with basic configurations of 1 × 3 or 1 × 4. If more channels are required, these basic elements can be daisy-chained to create, for example, an eight- or nine-channel switch. The losses of each step add up, which causes a decrease in throughput. However, the losses are much less compared to those with the shutter solution. Even an eight-channel switch has an efficiency of more than 50 percent, while an eight-channel shutter solution may have, at best, 10 percent.

The drawback is imperfect reproducibility: Short-term reproducibility values are in the range of 0.1 percent; long-term values are in the 1 to 2 percent range. And switching takes time. In this solution, as opposed to the shutter solution, the optical part not only has to be moved, but it also has to relax (stabilize again). Data rates higher than 20 spectra per second are tricky to achieve, and reproducibility may suffer at switching rates too fast, since the moved fiber might not relax enough.

The prices for optical multiplexers vary with the number of channels required, but the price per channel is lower than the cost of a good-quality spectrometer module.

Referencing

Measuring several channels more or less simultaneously offers the benefit of having a permanent reference channel available. Instead of using one channel to observe another measurement spot, it can be directed to, for example, a mirror (for reflectivity measurements) or a flat uncoated piece of glass (for transmission). Such referencing helps to eliminate drifts present in all light sources or originating from contamination of optics, depending on how the actual setup is executed. If only one spectrometer is used, time-dependent errors of the spectrometer module itself, such as sensitivity change due to a temperature increase, can be automatically corrected.

This optical measurement head is used in glass pane production.

A flashlamp has the advantage of low maintenance because the lifetime of a xenon bulb is quite high (billions of flashes), but a drawback is the high pulse-to-pulse variation in the percent regime. In combination with electronic multiplexing, one spectrometer can be permanently dedicated to observing the source and thus improving the stability of the measurement results by an order of magnitude. Sequentially operating an optical multiplexer cannot accomplish this; only simultaneous operation of multiple spectrometers will work. Such referencing can improve the signal quality by an order of magnitude.

Multiplexing can improve the amount and the accuracy of measurement data taken. However, there is no general overall solution. The various possibilities must be checked against individual needs to find the best approach. Features such as speed of measurement data acquisition, intensity level of the signal to be detected, accuracy demand, measurement principle, environmental conditions and, of course, cost have to be considered.

Meet the author

Gert Noll is the general manager at tec5USA in Plainview, N.Y. Tec5USA is the US subsidiary of tec5 AG in Germany.