Applied Science and Convergence Technology 2019; 28(4): 88-92
Published online July 31, 2019
Copyright © The Korean Vacuum Society.
Kwang Ki Kim, Jang Jae Lee, Si Jun Kim, Chul Hee Cho, Sung Wan Yoo, and Shin Jae You*
Department of Physics, Chungnam National University, Daejeon 34134, Republic of Korea
Correspondence to:E-mail: email@example.com
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-CommercialLicense (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution,and reproduction in any medium, provided the original work is properly cited.
Because radio frequency (RF) sensors typically experience interference between their capacitive detector and inductive detector, it becomes difficult for them to precisely measure voltage and current waveforms. We developed a high-precision RF sensor by using novel double walls (NDWs), which can minimize the interference. The geometrical construction variables of the NDWs are determined by analyzing the results from three-dimensional electro-magnetic simulation. The phase difference between voltage and current waveforms for the designed RF sensor is approximately 1.32° at a matched load, and that for an ENI probe (VIP1004) is 24.78°, as a result of which the developed RF sensor has better performance in preventing such interferences, compared with the other commercial sensor.
Keywords: RF sensor, Plasma, Voltage monitor, Current monitor, Novel double walls, 3D-EM simulation, Capacitive detector, Inductive detector
In many radio frequency (RF) systems, not all the output power of the RF generator reaches the load because of power losses in line and matcher [1–4]. Furthermore, upon using plasma chamber as a load, the power loss of the chamber is non-intuitive because plasma is a non- linear dielectric material . Therefore, to measure the correct power loss of the chamber, an RF sensor can be connected to the chamber for detecting voltage and current waveforms . By measuring the waveforms of the chamber as close as possible, excellent indication of plasma fabrication can be obtained. This, in turn, yields better control of the etching or deposition characteristics for a silicon wafer or other workpiece in the chamber.
Because of their various applications, RF sensors  have been widely studied for performing high-precision measurement of waveforms, impedance, and power. For performing high-precision measurement, it is extremely important to detect voltage and current waveforms independently. These waveforms can be obtained using capacitive detector and inductive detector, respectively. A capacitive detector detects voltage waveforms by using its capacitive electric field (CEF), and an inductive detector detects current waveforms by using its inductive electric field (IEF) . However, the inductive detector picks up the voltage both by the CEF and the IEF. Because of this behavior, the interference between both the detectors is practically unavoidable. To reduce the interference, Ibuki and Brown [9,10] used a metal wall between the detectors to design RF sensors. Nevertheless, these RF sensors still had the interferences, which resulted in imprecise measurement by the hardware itself.
In this paper, we developed a high-precision RF sensor, in two steps, by using novel double walls (NDWs). First, we analyzed voltages at specific locations inside of the RF sensor for minimizing the interference during three-dimensional electro-magnetic (3D-EM) simulation. The interference-minimizing process is detailed in the simulation study section. Second, we measured the phase differences between voltage and current waveforms by using the RF sensor at a matched load. The results were compared with those for the other commercial sensor.
In this section, the design of the RF sensor is discussed, following which a process for minimizing the interference between the voltage and current wave-forms is presented.
Figure 1 depicts the inside of the RF sensor, including the NDWs. As depicted in Fig. 1(a), the capacitive and inductive detectors are patterned on a single printed circuit board (PCB). The NDWs, which comprise three inner walls and four outer walls, penetrate the PCB. The geometrical construction variables of the NDWs are determined by D, H, and
Thereafter, it is necessary to minimize the interference for performing high-precision measurement of electrical parameters such as impedance and waveforms. The interference depends on the geometrical variables of the NDWs because the geometrical variables affect the capacitances that are formed among the core-conductor and the NDWs. In this regard, the interference can be minimized by changing the geometrical variables of the NDWs. To evaluate the interference, voltages V1, V2, V3, and V4 are indicated at specific locations, as shown in Fig. 1(a). The capacitive detector picks up voltage V2, which increases as voltage V1 decreases, because both the voltages are connected in series.
Meanwhile, the inductive detector picks up voltage V3, but this voltage comes from not only the IEF but also the CEF (that causes the interference). To distinguish between IEF and CEF in the inductive detector, voltage V4 is measured. Moreover, V4 is the voltage driven by the CEF in the inductive detector, because the IEF is minimum at this location. Therefore, the maximum condition of V3 over V4 means the inductive detector being affected only by the IEF.
In addition, we have to determine a less sensitive condition of performing voltage detection with respect to the geometrical variables of the NDWs. In the sensitive condition, when RF sensors are manufactured in large numbers, the degree of the interference for each sensor can be different because of manufacturing error. It results in poor repeatability of the RS sensor production.
In short, the voltage of the capacitive detector is increased as the ratio V2/V1 increases, and the interference of the inductive detector is decreased as the ratio V3/V4 increases. In addition, the less-sensitive condition must also be considered for minimizing the interference. We call these three conditions as the key conditions for minimizing the interference.
Figure 2 depicts voltages V1, V2, V3, and V4 with respect to the geometrical variables of the NDWs. In Fig. 2(a), voltage V1 increases upon increasing H because the capacitance formed by the capacitive detector and the NDWs increases. In contrast, voltage V2 decreases upon increasing H because the capacitance formed by the NDWs and the aluminum house (ground) decreases. Voltages V3 and V4 are less sensitive than H, meaning that H rarely affects the interference. Regardless of this, they become zero at H = 52 mm, because all the electric fields are shielded by the NDWs, as the ends of the NDWs reach the aluminum house. In Fig. 2(b), voltages V1 and V2 are varied faster than that in Fig. 2(a). It means the capacitance formed by the capacitive detector and the inner walls of the NDWs play an important role in detecting the voltage of the capacitive detector. In Fig. 2(c), voltages V1 and V2 are varied irregularly, but the variation is not so high. Voltages V3 and V4 are less sensitive versus
A four-step process is conducted to minimize the interference. First, voltages V1, V2, V3, and V4 are computed after setting arbitrary values for the geometrical variables of D and
Figure 3 depicts the voltage ratios V2/V1 and V3/V4, versus the geometrical variables H, D, and
Upon directly applying the output power of the RF generator to a load of impedance 50 Ω, the phase difference between the voltage and current waveforms at the load is ideally zero. Under this condition, if the interference of the RF sensor is minimized, the phase difference obtained from the RF sensor will be close to zero.
Figure 5 depicts a schematic of the experimental setup. The RF sensor is connected between an RF generator (NPG-1250A) and a matcher (Path Finder PF0113) by using N-type coaxial cables, and the sensor is also connected to an oscilloscope (TDS-3052B) by using SMA coaxial cables for measuring the waveforms. The oscilloscope is, in turn, connected to a PC by using the GPIB communication cable, and the probe condition of the oscilloscope is 50 Ω, as depicted in Fig. 8. RF power, ranging from 100 to 1000 W, at 13.56 MHz is applied to the matched capacitively coupled plasma chamber, and the chamber pressure is 5.4×10−6 Torr for preventing the generation of the dark discharge.
Figure 6 depicts the phase differences versus the power of the RF generator with respect to the RF sensor. From this figure, it can be observed that the averaged phase difference (between voltage and current waveforms) for the RF sensor is 1.32°, and that for ENI probe (VIP1004) is 24.78°. As a result of this experiment, the phase difference for the RF sensor is greatly close to that for the ideal case, compared with that of the ENI probe. On the basis of the electric-and magnetic-field distribution at 13.56 MHz in the inductive detector, which is not included in this paper, the electric field is shielded only within the NDWs. The CEF, however, is blocked by the NDWs. Therefore, low phase difference for the RF sensor results from successful blocking of the CEF by the NDWs in the inductive detector.
From this result, we can expect that the interference between the capacitive detector and the inductive detector is considerably reduced to be able to perform high-precision measurement of voltage and current waveforms, in contrast to that of the ENI probe.
The RF sensor is conventionally connected between the matcher and the plasma chamber. Under this configuration, the characteristic impedance of the RF sensor is as close to 50 Ω as possible. If the characteristic impedance exceeds 50 Ω, the RF generator can be damaged by the reflected power from the load. Therefore, we computed the transmission coefficient of the RF sensor in 3D-EM simulation as the information of the characteristic impedance .
Figure 7 depicts the transmission coefficient (S21) of the RF sensor and the error (in %) (=100×(1–S21)), versus frequency. In Fig. 8, S21 is very close to 1 under the frequency of 100 MHz with the error under 2 %. The result means that the characteristic impedance of the RF sensor is very close to 50 Ω.
Subsequently, the input impedance of the inductive detector was computed in 3D-EM simulation to confirm the bandwidth frequency of the RF sensor. The bandwidth depends on the self-resonance frequency (SRF) that comes from stray capacitance in the detector .
Figure 8 depicts the input impedance versus frequency for the inductive detector. In Fig. 8, the SRF is shown as 176 MHz, and the impedance is linearly proportional to the frequency under 100 MHz; the 0–100 MHz range can be defined as bandwidth because the stray capacitance in this range is negligible and hence can be ignored. Therefore, we expect that this inductive detector works suitably as an inductor in the bandwidth of 100 MHz. From this result, it can be noted that this RF sensor operates precisely with the bandwidth of 100 MHz.
As a result of this research, we developed a high-precision RF sensor by minimizing the interference between the two detectors by using NDWs in 3D-EM simulation. The bandwidth (100 MHz) and the characteristic impedance (50 Ω) were analyzed in the 3D-EM simulation as supplementary study. The RF sensor’s phase difference measured between the voltage and current waveforms was observed to be 1.32° for a matched load. This small value of phase difference means that the RF sensor is much more superior to another commercial sensor (VIP1004: 24.78°) for performing high-precision measurement of voltage and current waveforms.
This work is supported by Chungnam National University (CNU).