The problem of optical probing is the topic of this enlightening case analysis by Redding et al., of the Professor Dennis Prather group at the University of Delaware. They applied two different characterization techniques to observe the frequency-domain photoluminescence of a raised microdisk resonator. The use of a resonator does not restrict the validity of their conclusions to specific devices but provides a smart test bed that magnifies the measured effects. Resonators, especially at high-quality factors, are extremely sensitive to any surrounding modifications, this property making them widely employed in sensing applications.
A first way to perform optical probing is to collect photons that are “naturally” lost by the device, for instance, by using the far-field collection technique discussed by Redding et al., which detects the in-plane field radiated by the disk. Alternative techniques, such as the imaging of the out-of-plane scattered light [see for instance M. L. Cooper et al., Opt. Lett. 35(5), 784 (2010)], could also be employed with similar results. By definition, these approaches are nonperturbative. Their main limitation is that information is carried only about those photons that are radiated out along preferential directions. What happens if this photon number is small compared with the number of photons absorbed into the device or escaping in other directions? The measurement becomes noisy and can even fail. This might sound a bit surprising, but actually measurement techniques based on the collection of the radiated (or scattered) field work much better on “bad” devices with high radiation loss.
In the case of devices with small radiation loss, we are forced to spill out a higher number of photons by means of some coupling mechanism. Examples of these approaches are the tapered fiber collection technique (used by the authors) and scanning near-field optical microscopy. In both cases, the evanescent field at the boundary of the DUT is partially caught by a tapered fiber probe placed at a proper distance from the DUT. These are by definition perturbative techniques. In the experiment shown in this work, the loading effect given by the fiber probe results in the noticeable broadening of the emission linewidths when the fiber is placed closer to the disks. Here a fundamental constraint exists between the sensitivity of the measurement apparatus (how many photons are required at the receiver) and the DUT sensitivity, setting the maximum and the minimum distances, respectively, at which the fiber probe can be placed. If the two ranges do not overlap, that is, if the DUT is more sensitive than the measurement apparatus, we stand no chance of making an unperturbed measurement.
The main conclusion of this work is that, to date, the problem of optical probing is still waiting for an answer. And the general feeling is that the need for smart solutions is growing, because both the integration scale of optical devices and the number of optical functions that can be embedded in a single chip are increasing. Fast and accurate optical probing is essential for efficient test-on-wafer diagnostics, fault detection, postfabrication trimming, real-time monitoring, and adaptive tuning of the devices. In view of this, accessing information locally and with no device manipulation is the fundamental step toward optical signal processing and system-on-chip applications.
You must log in to add comments.