VI Probes
Introduction:
Measurement of the current, voltage and phase are perhaps the most important aspect to any RF plasma system network. From these we can learn about the power delivered to the plasma, and load impedance which will directly affect plasma processes. VI probes are an effective solution to this, as they can measure the voltage, current and phase at any point in an RF network, from which the power and complex impedance are calculated. This article will explain how they work and how they can be used to monitor plasma processes.
Section 1: Understanding VI Probes:
As previously mentioned, VI probes measure the voltage, current and phase of an RF signal. They do this through remote current and voltage pick-ups; a current loop is used to detect the magnetic field and measure the current, and a capacitor is used to detect the electric fields and measure the voltage as shown in the image below. This is done in a calibrated box to get read absolute values. The phase between these signals can then be determined. RF power at any point can be calculated by the following equation, where P is power, I is current, V is voltage and 𝜃 is phase.
𝑃=𝑉 𝐼 𝐶𝑜𝑠(𝜃)
From these measurements, a wide variety of parameters can be calculated including the forward and reflected power, the energy deposited, the VSWR, and complex impedance.
Figure 1 Block diagram of V-I probe
VI probes exhibit many qualities that make them better suited for RF power measurements over other methods such as directional couplers, which involve the coupling of the RF signal to transmission lines for highly directional measurement of forward and reflect powers. Firstly, for VI probes measurements can be performed even outside of the 50 Ohm region between the generator and the matchbox. They can monitor several frequencies simultaneously, measuring the power and impedance of each frequency individually. In comparison a directional coupler will have weak performance when harmonic content is present. The remote pick-ups technique is used by VI probes which means power measurements can be precise from milliwatts to tens of thousands of Watts, by controlling the pick-up distance and sensitivity. Typically for directional couplers there is power loss at each port used as a few watts are absorbed. Therefore, when calibrating reflected power for generators, a directional coupler may report 0 Watts, but there is actually 1 or 2 Watts present.
A challenge with VI probes however is the crosstalk between voltage and current pick-ups, which need to be accounted for in the calibration. Furthermore, the phase accuracy is critical when the phase is near ± 90 degrees as a small error in phase at this value will result a large error in the power measurement because of the 𝐶𝑜𝑠(𝜃) term.
Section 2: Impedans Octiv VI Probes
For accurate measurements of voltage, current and phase in an RF circuit, Impedans offer advanced Octiv sensors for a wide variety of applications. These VI probes undergo an impedance calibration that is traceable to NIST standards, wherein the output of the sensor is periodically connected to open, short and load standards for repeatability and verification of measurements. Calibration is also done up to 80℃ in a temperature-humidity controlled chamber to ensure that when no inputs change, the output remains constant. Power measurements are calibrated using a calorimeter for fundamental and harmonic frequencies plus a ±10% band around the central frequencies. When all calibrations are complete, verification tests are done against the calorimeter and a Vector Network Analyser (VNA) to ensure calibration transfer was successful. The resulting sensor combines the impedance measurement accuracy of a VNA and the power measurement accuracy of a calorimeter in a compact easy to use device.
Figure 2 Impedans Octiv Suite 2.0 for RF measurements
Standard Octiv sensors have a 12kW specification however, custom form factor solutions are available for up to 90kW, achieving a power uncertainty of 1% maintained to a VSWR of 6:1. Similarly maximum current and voltage specifications can be increased from 3kV pk-pk and 25A (standard) to 7kV and 70A (custom). Below is a plot of Impedance measurements made by the Octiv at different regions of the smith chart, compared to an offline VNA at 2MHz.
Figure 3 Impedance measurement by Octiv at 2 MHz
The Octiv features a typical report ratio of 10/second, with up to 500/second available as a standard increasing to 33000/second with special Alfven sensor. These fast response times allows the Octiv to resolve pulsed signals, plot the shape of applied RF pulses and measure the stability of consecutive pulses over time. The Octiv will automatically detect RF pulsing and begin to report the pulse frequency and duty cycle in real time. In a specific pulse mode, it will also take an integral over the pulse on time to accurately measure average power, voltage, current and impedance while pulsing. Therefore, if you are operating at 1kW continuous wave, then turn on. A 1kHz pulse at 50% duty cycle, the Octiv will still report 1kW power plus the pulse frequency and duty cycle automatically.
Section 3: Applications and Benefits of VI Probes
An RF power network will typically have 5 main sections, namely the RF generator, the 50 Ohm coaxial cable, the matching box, the non 50 Ohm connector and the plasma reactor (load). Every part of this network has an impact on the plasma’s behaviour and in return the plasma has an impact on every part of this network. For instance, both connectors will be worn down over time and result in losses. For this reason, the generator must be calibrated to compensate, which can be done by measuring the power delivered at the entrance of the matchbox. This will measure the power that the generator is actually outputting for the plasma process. Measuring the forward, reflected and delivered power in the 50 Ohm region is vital for monitoring process performance it will assess the efficiency of power transfer and help identify any impedance mismatches. An increase in reflected power can also indicate the presence of faults, defects, or malfunctions in the transmission line, connectors, or components.
A key application for VI probes regarding the RF network is the characterisation and control of the RF match box. This can be done easily by connecting an RF power supply to the output of the match box and a 50 Ohm load to the input, then connecting VI probes in series either side of the match. The match box is then cycled through all its impedance settings to measure the impedance range of the match. The plasma impedance that’s measured during a process must lie within this range for efficient power delivery. This is especially vital for new processes on old tools. Matching efficiency as a function of impedance phase of the load can be measured using the two VI probes on either side of the matchbox to fully characterise matchbox performance over its range. Below is example plots made using Impedans Octivs for the calculation a match boxes range (left) and efficiency (right) in figure 4. 
Figure 4 Impedance range and efficiency of a matchbox as measured by Octiv VI Probe.
A plasma is essentially a conducting gas which can be modelled as an equivalent circuit of resistors, capacitors and inductors in this RF delivery network. In the 50 Ohm region, after the match box, the impedance is dominated by the plasma. From the point of view of a VI probe placed here, the sum of the impedance is measured as the total plasma resistance, which is very sensitive to these resistor, capacitor and inductor values.
The capacitance comes from the plasma sheath, which is the small gap between the plasma and the walls or substrate. These values will change if the wall conductivity changes, therefore it will be dependent on chamber seasoning or cleaning. It will also change with sheath thickness, which is related to the plasma density and hence from measuring the capacitance you will be able to see fundamental changes in the plasma itself. It will even change if a wafer substrate in a process is different or misplaced, or when there is charge build up on a wafer. From this incorrect RF biases or asymmetric charge build up can be seen before arcing happens and loss of product occurs.
The resistance of the circuit is due to electron collisions in the plasma as they move through the chamber to complete the circuit. Low pressures have large mean free paths, so electrons may travel millimetres to centimetres without collisions. However, they are travelling very fast, so collisions still happen often therefore the resistance will change with gas pressure changes. Different gases have different collisional cross sections with electrons so gas contamination can be detected, as well as air leak and gas flow issues. This also includes chamber clean and etch endpoint detection: as a plasma cleans or etches a material, this material will be released into the chamber. When the material is gone, the plasma impedance changes, which can be measured.
The shape of the RF waveform in the non 50 Ohm region will be a combination of all these effects; each fault will contribute a component. An Octiv VI probe will sample the RF voltage and current waveform, then perform a Fourier Transform, to convert the wave from time and amplitude to frequency and amplitude as shown in the plots below. These frequencies are called the plasma harmonics, and plasma faults will cause unique patterns in this spectrum. For example, harmonic 3 might change with pressure and harmonic 6 might change with wall condition etc. The idea for monitoring these harmonics is that if this spectrum is the same, then the plasma process is the same.
Figure 5 Left: RF current and voltage waveform, Right: RF harmonic spectra of waveforms
As mentioned before, key applications of harmonic monitoring are clean and etch endpoint detections. For a wall cleaning process on a deposition tool, the power delivered remains constant for the entire clean duration. Monitoring the RF harmonics can reveal the change in the plasma state (and thus chamber condition) between the start and end of the clean. For etch endpoint detection, Optical Emission Spectroscopy (OES) is commonly used. However, this technique struggles to monitor endpoint for small features, due to the small amount of material that enters the plasma as new layers are etched. The data is often too noisy below 5% open area ratio, and below 1% is considered not possible with OES. However, the harmonic method has been shown to be able to detect endpoint below 1% open area ratio in the field, as shown in below study by Jang et al (2013)1. Clear etch endpoint detections are essential for process optimisation, and fine control over etch depth profiles as the plasma won’t over or under etch your material.
Figure 6 Etch end point detection using optical and RF harmonic methods in low open area processes
Conclusion:
To conclude, VI probes have emerged as essential tools, providing real-time measurements of voltage, current, and phase relationships in various plasma environments. These measurements reveal the complexities of plasma behaviour, empowering researchers and industries to refine processes, enhance product quality, and bolster operational efficiency. Impedans, a leader in plasma diagnostics, offers state-of-the-art VI probe solutions that equip industries and researchers to navigate the intricacies of plasma systems with confidence. For more information on VI probes and how they can be used for your process, please navigate to the Impedans website or get in touch with one of our plasma experts.
References:
1Jang, H., Nam, J., Kim, C.-K. and Chae, H. (2013), Real-Time Endpoint Detection of Small Exposed Area SiO2 Films in Plasma Etching Using Plasma Impedance Monitoring with Modified Principal Component Analysis. Plasma Process. Polym., 10: 850-856. https://doi.org/10.1002/ppap.201300030
