edfas.org 29 ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 25 NO. 4 the incident beam energy. The various processes leading to backscattered and secondary electron emission are detailed by Lin and Joy,[1] a compilation of over a century of secondary electron data for various materials is provided by Joy,[2] and Thong[3] provides secondary and backscatter data for several materials. Lin and Joy conclude that higher-energy primary electrons create more secondary electrons as they scatter within the DUT, but they also travel deeper into the DUT before losing their energy, and the chance of a secondary electron escaping from the DUT drops exponentially with distance from the DUT surface. As a result, for most materials, there is a peak in secondary electron yield—defined as the number of secondary electrons emitted by the DUT for each primary electron incident on the DUT—for incident beam energies of about 500 eV. Below 500 eV, fewer secondaries are created within the DUT, whereas above 500 eV more secondaries are created, but at greater depths where it is harder to escape the DUT. Lin and Joy estimate the maximum secondary electron yield for Si, Al, and Cu (three materials common in modern ICs) to be 0.89, 2.00, and 1.53 at primary beam energies near 450 eV, while calculated escape depths are 2.7, 1.7, and 0.6 nm, respectively.[1] Secondary electrons having low-eV energies and only needing to travel nm distances to escape from the DUT are generated within femtoseconds of the primary arrival time, thus capturing the state of the DUT within nanometers and picoseconds of the primary electron’s arrival position and time. Therefore, techniques such as voltage contrast, which derive from secondary electron emission, work equally well from DC through GHz speeds. A secondary electron yield of 2, as for Al under bombardment by 450-eV primary electrons, implies two secondary electrons (on average) leave an Al target for every primary electron that strikes the target. However, total electron yield, including backscattered electrons[3] will exceed 2. If the Al DUT is not maintained at ground potential, it will charge to a net positive potential while being bombarded by 450-eV electrons from the SEM. VOLTAGE CONTRAST Thong explains the mechanism for voltage contrast, which is a variation in secondary emission for changing DUT voltages.[3] While secondary-electron emission itself is a stochastic process governed by the physical properties of materials, it turns out that there are many factors that can cause a voltage-contrast effect to appear in SEM images. This section describes several such factors, and provides a baseline, determined empirically, from which VC sensitivity predictions can be made. The dominant mode of SEM imaging derives its signal from secondary electrons emitted from the DUT. The emission spectra of secondary electrons can be measured with an electron spectrometer and peaks for secondary electron energies in the 1 to 3 eV range.[3] By building some form of electron spectrometer into the SEM’s secondaryelectron detection path, it is possible to detect secondary electrons above certain cutoff energies while rejecting those of lower energy. Such spectrometers form the basis for all voltage-contrast imaging. Two examples help clarify this issue. In the first example, a retarding potential in the secondary electron detection path rejects all electrons below 100 eV. In this case, all secondary electrons which have energies below 50 eV will be rejected, and only backscattered electrons will contribute to the image. Back-scattered SEM images are formed in such a manner. In a second example, a weak retarding “mirror” potential may be created within the SEM column that rejects certain low-energy secondary electrons while allowing high-energy backscattered electrons to pass through. This “mirror” potential slows mid-energy secondary electrons long enough for them to be steered into a detector. Changing this “mirror” potential changes the energy threshold at which secondary electrons are detected. For normal SEM operation, it might be desirable to collect as many secondary electrons as possible to form the best SEM image, but this may not be the case for voltage-contrast imaging. As described in Thong secondary electrons emitted from a DUT at non-zero potential will experience a shift in energy as they enter a grounded SEM column, and an electron discriminator will now pass fewer (or more) secondary electrons as the DUT voltage increases (or decreases).[3] This is the primary mechanism behind VC imaging; in the presence of some type of energy discriminator that causes only mid-energy secondary electrons to contribute to the SEM detection signal, high DUT voltages lead to dark SEM images, while low DUT voltages lead to bright SEM images. It is up to the SEM designer and user to optimize the energy discriminator to maximize the VC signal for any given secondary-electron emission spectra. Unlike secondary electron emission, voltage contrast is not determined entirely by physics. Several environmental factors besides DUT voltage alone can influence the number of secondary electrons that are detected, and some examples follow. 1) For IC circuitry, potentials on surfaces near the probe location can recapture or steer low-energy secondary electrons toward or away from (continued on page 32)
RkJQdWJsaXNoZXIy MTMyMzg5NA==