| Electron-Beam Probing (eBP) [1, 2] is a specialized technique used in fault isolation and failure analysis in semiconductor devices. The E-beam prober system uses voltage contrast to detect surface potentials on a device under test (DUT). By focusing a scanning electron microscope (SEM) on the DUT's surface, it generates secondary electrons that create images showing surface topography and voltage contrast. The brightness of the surface in these images varies with the applied bias, indicating changes in surface potential. The ability to detect voltage differences is determined by a relationship involving the SEM's spectrometer constant, CC, beam current I, and integration time T:
 ------------------------------------------------ [0178a]
where,
ΔVRMS
is the root mean square (RMS) value of a voltage fluctuation.
I is the beam current.
T is the integration time.
The eBP system (see eBP setup) measures voltage differences using a scanning electron microscope, and the ability to discern these differences is quantified by Equation 0178a. Such a system supports two modes:
- Continuous electron-beam current for probing and imaging. In this case, the system maintains a steady, uninterrupted electron beam during probing so that the electron beam provides a consistent stream of electrons that interact with the surface of the sample to generate signals (such as secondary electrons). This is particularly important for detecting defects, performing fault localization, and ensuring high-quality imaging, especially at very small scales like those required for advanced semiconductor nodes (such as <5 nm).
- Pulsed electron-beam mode for higher-frequency signal measurement. In this mode, the system uses a blanker to supports three speeds of blanking, resulting in
various pulse-width options, or bandwidths. The fastest
blanking bandwidth can be at least 2-GHz. [3] This is done to investigate the timing characteristics of electronic devices, especially those operating at very high frequencies, like those in the Gigahertz range:
- Timing Analysis and Design Debug: In advanced semiconductor devices, timing issues are important. Devices operating at Gigahertz frequencies require precise timing to function correctly, and any deviation can result in malfunction or decreased performance. By pulsing the electron beam at these high frequencies, the system can measure the behavior of signals within the device, such as clock signals or data paths, and identify timing-related problems.
- Frequency Spectrum Measurement: A pulsed electron beam allows for the measurement of frequency spectra, which is the distribution of signal frequencies present in a system. This helps in analyzing how different parts of the circuit behave at various frequencies, making it possible to identify issues like signal integrity problems or clock jitter.
- Design Debug: At such high operating speeds, it’s essential to capture fast transient events that may not be visible with a continuous electron beam. The pulsed beam, combined with its high bandwidth, enables the system to capture and analyze these fast events, providing valuable data for debugging design flaws in high-speed circuits.
Figure 0178a shows the root mean square (RMS) value of a voltage fluctuation depending on beam current and integration time.
| Figure 0178a. Root mean square (RMS) value of a voltage fluctuation depending on beam current and integration time. |
In the measurement, electrical signal can be measured or analyzed from the device under test (DUT) using the electron beam probing system. That is, the time-domain signal s(t) is captured at a specific point on the DUT. This signal represents the voltage or current variations over time at that particular location. This is, the signal s(t) is associated with a pixel on the image when the electron beam scans the device under test (DUT) in an SEM system.
Then, the I and Q components can be derived from the signal using the discrete Fourier transform as described below:
Then,
the amplitude is calculated by combining the I and Q components as follows:
 -------------------------------------------------- [0178d]
This equation gives the magnitude of the signal at each pixel, which can then be visualized across multiple pixels to produce an amplitude image.
On the other hand, the phase can be calculated using the arctangent function:
 -------------------------------------------------- [0178e]
The atan2(Q, I) function returns the phase angle of the complex number formed by I and Q, taking into account the correct quadrant. This phase value can be visualized across multiple pixels to produce a relative phase image.
Figure 0178b shows the calculated amplitude and phase as well as the simple, original sinusoidal signal, s(t), which represents the electrical signal that is measured from the device under test (DUT) using the electron beam probing system.
| Figure 0178b. Calculated amplitude and phase as well as the simple, original sinusoidal signal, s(t). |
The ebeam probing can be used to detect dynamic signals in metal structures. That is, the electron beam technology can observe changes in electrical signals (such as voltage fluctuations) in real-time within metal interconnects or circuits.
Ebeam probing is commonly used in semiconductor failure analysis to observe and measure electrical activity at the nanometer scale, offering the ability to detect transient or dynamic behaviors in circuits that other techniques might not capture as precisely. This is especially useful in identifying issues in high-speed or high-frequency circuits where signals change rapidly.
There are a few key vendors that provide eBP systems or related technologies:
- Thermo Fisher Scientific - While they are more known for their electron microscopes, they also provide systems and tools that can be used for electron beam probing in semiconductor analysis.
- Advantest - They offer solutions for e-beam probing, particularly for semiconductor fault isolation. Advantest's E3650 system is an example of an e-beam test system used for high-precision measurements.
- Kleindiek Nanotechnik - This company specializes in nanoprobing tools and has systems that can be integrated with electron microscopes for e-beam probing applications.
- Zyvex Labs - Known for their work in nanotechnology, they provide tools that can be used in conjunction with e-beam systems for probing at the nanoscale.
In signal detection and failure analysis, duty cycle plays a critical role in determining the effectiveness of measurement techniques like EVI and LVP. The duty cycle refers to the percentage of time a signal is active ("on" state) within a given period, with larger duty cycles generally providing better signal detectability. However, the relationship between duty cycle and Signal-to-Noise Ratio (SNR) is not linear. While intermediate duty cycles, such as 50%, often yield the highest SNR by balancing the active and inactive states of the signal, extremely large duty cycles can lead to signal saturation and higher exposure to background noise. This reduces the number of useful signal transitions that can be captured, limiting the system's ability to distinguish between signal variations. On the other hand, very low duty cycles make signal detection challenging due to the reduced active period of the signal, leading to a weaker SNR. The ideal duty cycle for optimizing SNR in signal detection typically falls between these extremes, with techniques like the two-point transformation being developed to improve SNR even for low-duty-cycle signals.
eBP can provide the following applications:
- Direct Probing of Metal Interconnects or Active Transistors:
Unlike optical techniques (LVx) that struggle with photon penetration through dense metal layers, eBP can probe the metal interconnects and even the transistors directly. This is crucial for analyzing advanced transistor architectures.
- Voltage Contrast:
eBP exploits the voltage contrast phenomenon to detect surface potentials in a device under test (DUT). When primary electrons interact with the DUT, they generate secondary electrons, and the strength of these secondary electrons is modulated based on the switching frequency of the device. This provides insight into the electrical activity at various points in the circuit.
- Electron Voltage Imaging (EVI):
EVI uses a Discrete Fourier Transform (DFT) algorithm to map specific frequencies, such as system clocks, across the field of view. This allows engineers to quickly identify the electrical activity of the device and is particularly useful in diagnosing scan chain integrity issues.
- Electron Voltage Probing (EVP):
EVP complements EVI by providing real-time waveform recordings. These waveforms are sampled to trace specific signals that might be of interest in debugging or analyzing failures in the circuitry.
- Application to Low Duty Cycle Signals:
eBP is able to detect signals with low duty cycles (less than 10%), particularly for circuits designed to mitigate power consumption through techniques such as clock gating. For example, eBP can validate timing and debug signals in low duty cycle architectures used in modern GPUs or during power throttling.
- Scanning for Fast-Pulsed Signals:
The system is capable of capturing high-speed signals, even in the sub-nanosecond range (e.g., a pulse width of 350ps). This makes eBP suitable for analyzing fast, transient electrical events that are difficult to capture with other methods.
- Phase-Locked Detection for Signal Analysis:
eBP can synchronize with the device's clock to capture signal variations over time. This makes it possible to trace signals that have specific frequency or phase relationships to the device's clock, improving the accuracy of failure analysis.
eBP's applications focus on overcoming the limitations of optical techniques, providing in-depth analysis of electrical activity at the metal and transistor levels, detecting low duty cycle signals, and probing fast-pulsed signals in advanced semiconductor architectures.
However, there are several limitations in eBP technique:
- Time-Consuming Measurements:
eBP, particularly EVP, can be a slow process. Measurements often take several minutes, especially when optimizing the signal and probe location. This makes it less efficient than some optical methods like LVx, which can provide faster feedback.
- Dependence on Signal Duty Cycle:
While eBP is effective at detecting low duty cycle signals, its performance can still be significantly impacted by the duty cycle. For EVI, the SNR decreases as the duty cycle deviates from 50%. This makes it harder to detect signals with very low duty cycles unless additional measures, like a two-point transformation method, are used.
- Signal Periodicity Requirements:
eBP relies heavily on detecting repetitive signals, especially in the EVI mode. Signals with consistent clock patterns, such as those used in scan chain integrity analysis, are ideal for this technique. However, irregular or non-repetitive signals may not be effectively analyzed using standard eBP techniques, limiting its versatility in capturing random or one-time events.
- Complexity in Phase-Locking:
To improve measurement accuracy, the eBP system often needs to be phase-locked to the device under test (DUT). Achieving this synchronization can be technically challenging, especially in asynchronous systems where the device and the eBP system are not operating on the same clock. Without phase-locking, the signal-to-noise ratio can be reduced due to noise in the measurement process.
- Secondary Electron Detector Bandwidth:
The performance of eBP can be limited by the bandwidth of the secondary electron detector, which affects the temporal resolution of the system. For example, the electron pulses can be constrained by the detector's bandwidth, limiting the effectiveness of high-speed measurements.
- Inability to Retain Frequency Information in Some Methods:
The two-point transformation method used in eBP provides amplitude information of the signal but loses frequency details. This is a trade-off for faster data acquisition and improved signal-to-noise ratio, but it limits the ability to analyze the frequency characteristics of the signal, which can be critical in certain failure analysis scenarios.
- Sample Preparation Complexity:
While eBP can directly probe metal interconnects and transistors, this often requires complex sample preparation using Focused Ion Beam (FIB) techniques. FIB preparation can be time-consuming and may introduce artifacts that affect the accuracy of the analysis.
[1] Thong, John T. L. (Editor), Electron Beam Testing Technology, Springer, doi.org/10.1007/978-1-4899-1522-1, New York, 1993.
[2] W.T. Lee, Engineering a Device for Electron Beam Probing, IEEE Design and Test of Computers, Pg 36, 1989.
[3] Neel Leslie, James Vickers, Blake Freeman, Seema Samani, Prasad Sabbineni, Praveen Vedagarbha, 2GHz Contactless Electron Beam Probing, ISTFATM 2022: Conference Proceedings from the 48th International, https://doi.org/10.31399/asm.cp.istfa2022p0125, 2022.
|
-------------------------------------------------- [0178b] 
-------------------------------------------------- [0178c]