Recently, the development of electronic products has been accelerated for various applications. Accordingly, electronic products are becoming smaller and faster, for better performance. Such advanced electronics require short wiring lengths and narrow line widths in printed circuit boards (PCBs). However, narrow lines and spacing can lead to serious defects, such as open or short circuits from copper-plating deviation [1]. Typically, copper-plating defects are caused by several factors, including electrolytes, plating conditions, external forces, and foreign materials. These types of copper-plating defects can be resolved during the electroplating processing. However, the resulting problems caused by the copper-plating deviation cannot be easily addressed after completion of the electroplating manufacturing process.

Currently, several instruments are used to measure copper-plating thickness, including the X-ray fluorescence spectrometer (XRF), eddy-current coating-thickness tester, electromagnetic coating-thickness tester, coulometric coating-thickness tester, optical microscopy (OM), and scanning electron microscopy (SEM) [2–5]. The principles of XRF are as follows: First, when a copper-containing sample is irradiated by X-rays, it emits its own fluorescent X-rays based on the amount of copper present in the sample. Then the XRF can detect the intensity of the fluorescent X-rays, which determines the plating thickness [2]. However, it is difficult to perform in situ measurements via XRF. Furthermore, XRF is expensive, with a large scale. The eddy-current coating-thickness tester uses the following principle: When an electrode with high-frequency current flow is placed on a sample, an eddy current is generated along its surface. Given that the intensity of the eddy current varies according to the plating thickness, it is measured to determine the plating thickness [3]. However, in situ measurements are difficult to perform with an eddy-current coating-thickness tester. Moreover, this type of contact method can damage the sample during measurement. An electromagnetic coating-thickness tester is a noncontact method. When an electrode approaches a sample, the electromotive force between the two changes, and the plating thickness can be inferred from the difference in this electromotive force [4]. However, an electromagnetic coating-thickness tester also cannot perform in situ measurements effectively. Moreover, its resolution is insufficient for applications involving practical processes. A coulometric coating-thickness tester can measure the plating thickness of a dissolved sample; However, it is not suitable for practical applications, as it is a destructive method that requires additional equipment [5]. OM and SEM are also destructive methods for measuring plating thickness via cross-sectional photography, and furthermore excessive time and effort are required for the prerequisite process of cutting the sample to obtain a thin cross-sectional face.

Therefore, in practical processing applications it is extremely difficult to measure plating thickness via the conventional equipment mentioned above, because of various limitations, including sample contact, sample destruction, expense, and the time-consuming nature of the measurement methods, along with sensing problems. In the practical electroplating process of a PCB, the overall surface is on the order of several square meters. Therefore, multiple platinum-mesh electrodes are required for a rapid and wide electroplating process. For the regulation of plating deposition to satisfy the overall plating-thickness specification, in situ precision sensing over multiple positions across the entire surface of the plating sample is highly desirable. Therefore, in this study we present a novel method for measuring the plating thickness in situ via optical Fourier-domain reflectometry (OFDR) system technology.

The OFDR system is similar to optical coherence tomography (OCT) in principle. It uses a wavelength-swept laser to scan the lasing wavelengths repeatedly at regular intervals, leading to optical interference between the partially reflected light from the sample surface and the remaining light passing along a given optical-path difference (OPD) of the interferometer. Given that the information regarding the distance is determined from this interference signal [6–8], the height variation in the sample surface can be directly analyzed at the given directional position. Typically, even tiny OPD can lead to phase differences in optical interference; Hence OFDR is capable of high-resolution distance at nanometer-scale resolution. Additionally, provided that a high-speed measurement of 100 kHz or higher is possible, the plating thickness is measured in real time, in a noncontact and nondestructive manner [9, 10].

Based on these principles, OCT and OFDR systems are widely used in various fields requiring distance measurement [11–14]. In practical applications, these systems are also widely used in biomedicine, including the measurement of corneal or tissue position during surgery [15, 16].

The optical-fiber probes for OFDR or OCT have multiple advantages, including small size, light weight, and high resistance to corrosive chemicals. The flat-tip designs based on the common-path (CP) interferometer feature simple schematics without a separate reference arm, convenient interchangeability, low cost, and robustness to external vibrations [17–20].

However, since the output beam of the flat-tip design has a divergent beam profile, signal loss increases with the distance between the fiber tip and sample. Meanwhile, if the distance between the sample and the fiber tip is too short, the intensity of the light reflected from the sample may exceed the saturation limit of the photodetector. Therefore, an optimal distance between sample and fiber tip must be carefully maintained for the maximum signal-to-noise ratio (SNR).

In this study, we demonstrate two OPD-regulated OFDR probes that can move vertically and regulate the preset distance from the sample via feedback control, by combining a linear piezo motor and an OFDR system. We implement the OPD-regulated system on each probe for two reasons: Optimal distance setting before the plating-thickness measurement, and optical-fiber protection to avoid bumping into the sample during motion. The proposed OPD-regulated sensing module enables accurate and repeatable distance regulation.

Furthermore, the OPD-regulated OFDR probe can prevent damage to the fragile fiber tip or the sample due to their unexpected contact during probe movement. The wide stroke of the piezo linear motor (30 mm) allows centimeter-range vertical movement with high speed and precision.

We also propose a real-time plating-thickness monitoring system with nanometer resolution based on the phase variation of the interference signal. We successfully demonstrate real-time uniformization of the plating thickness during electroplating. The copper-plating deviation was controlled by comparing the plating thicknesses at multiple positions on the sample.

2.1. Experimental Setup

Figure 1 shows the configuration of the proposed OFDR system, with two identical OPD-regulated sensing probes A and B to measure the plating thickness at multiple sample positions. Multiple OFDR sensors must be applied to measure the thickness deviation of electroplating depending on the locations upon a sample. We use two functionally and operationally identical probes in this experiment. The light source is a wavelength-swept laser (HSL-10/20; Santec, Aichi, Japan) with central wavelength of 1,310 nm, a −3 dB bandwidth of 80 nm, a coherence length of up to 18 mm, peak output power of approximately 50 mW, and a sweep rate of 100 kHz. The laser output is divided into the two paths of the sensing probes via a 50:50 1 × 2 optical coupler (OC) (TW1300R5A1; Thorlabs, NJ, USA). Each divided laser beam is guided into the first port of the circulator (CIR) (CIR1310; Thorlabs), with the second port attached to a sensing probe. Using the CP structure, interference is generated between the light reflected from the end-tip surface of the optical fiber (SMF28; Corning, NY, USA) and that reflected from the sample surface by OPD [21, 22]. Without an antireflection coating on the optical fiber, Fresnel reflection (~3.5% at the glass–air interface) occurs from the flat surface of the silica fiber’s end tip, providing the reference light. Each beam reflected from the sample surface and the reference surface creates an interference signal with a frequency proportional to the OPD between the two surfaces. For a single reflector at a distance d from the reference surface and a sufficiently long coherence length (without signal roll-off depending on d), the signal current i_sig can be simply expressed as [9]

Figure 1. Schematic of OPD-regulated OFDR and distance actuation using linear piezo-motor feedback. OPD, optical-path difference; OFDR, optical Fourier-domain reflectometry; OC, optical coupler; CIR, circulator; PD, photodetector; RT, rigid tube; DR, driving rod; OF, optical fiber; PM, piezo motor; R_r, reflection of the reference surface; R_s, reflection of the sample surface; d, distance between R_r and R_s.

(1)$$i_{\text{sig}}\left(t\right)=\frac{\eta q}{h\nu }\text{2}\sqrt{I_r{I}_s}\text{cos}\left(\text{2}nk\left(t\right)d+\phi \left(z_{\text{0}}\right)\right)$$

where η is the detector sensitivity, q is the quantum of electric charge, hν is the single-photon energy, I_r is the optical intensity reflected from the reference surface, and I_s is the optical intensity reflected from the sample. Here k(t) = 2π/λ(t) is the wave number varying in time by the wavelength-swept laser; z₀ = 2∙n∙d denotes the OPD, which is twice the beam path length from the sample and the reference; And φ(z₀) denotes the phase of i_sig at z₀. From the Fourier transform of i_sig, d and δd are easily obtained: d is the distance between the end tip of the silica fiber and the sample surface, and δd is the time-varying and continuous change in the distance at sub-resolution scale. The positive-side frequency-domain spectrum of i_sig has a peak value at z₀ (d = z₀/2n). The phase-based distance variation δd can be calculated from φ(z₀), and is expressed as

(2)$$\begin{array}{c}\delta d=\frac{{\lambda }_c}{\text{4}\pi \cdot n}\cdot \delta \phi \left(z_{\text{0}}\right)\end{array}$$

where λ_c is the central wavelength of the wavelength-swept laser, and n is the refractive index of the medium.

The third port of CIR is connected to a photodetector (PD) (PDB410; Thorlabs), which converts the optical interference from the sensing probe into an electrical signal. When the A-line trigger from the laser is on, a digitizer (ATS9350; Alazar Tech, Quebec, Canada) with a sampling rate of up to 500 MSPS starts acquiring the interferogram signal of a single A-line. For linear response to change in distance, as well as optimal sensitivity, the raw interferogram signal is transformed linearly in the wave-number domain using the interpolation method. Afterward, windowing, zero padding, and fast Fourier transform are performed on the interpolated A-line data. These processes produce complex A-line data (with real and imaginary parts), with information displayed on the frequency spectrum. Finally, the magnitude and phase of the complex data are used to calculate the distance d for the OPD-regulated system, and the distance variation from the phase information for plating-thickness measurement at a nanometer scale, respectively [23, 24].

For OPD regulation, the sensing-probe module consists of an optical fiber integrated with a small rigid tube through the central hole, combined with a driving rod (DR); A linear piezo motor (PM) (LL1011A; Piezo Motor, Uppsala, Sweden) creating the linear movement of the DR, with a resolution of less than 1 nm, maximum speed of 15 mm/s, and stroke length of 30 mm; Along with an external housing to protect the module. The PM is coupled securely inside the housing and fixed to a stable external holder.

According to the measured OPD from the OFDR system, the optical fiber’s position is vertically shifted using the motion controller (PMD101; Piezo Motor) to regulate the distance between the sample surface and the tip of the optical fiber in real time. The magnitude of the position movement is calculated through a feedback-loop system based on optimized proportional-derivative control. For example, if the end of the fiber is closer than the preset distance, the linear PM moves backward. In the opposite case, the motor moves forward to regulate the preset distance, thereby preventing contact with the sample.

2.2. Results of OPD-regulated OFDR

Figure 2 shows the schematic of the experiment, and results for the OPD-regulated OFDR system. Figure 2(a) shows the schematic of the OPD-regulation test on a multi-grooved metal sample with 2-mm intervals and a 0.6-mm dip. The OPD-regulated OFDR system is moved horizontally at a speed of 5 mm/s using a translation stage (MLS203; Thorlabs). As shown on the left in Fig. 2(b), when the OPD regulation is off, the distance of the grooved sample from the fiber tip end is calculated in real time for the first second, while the height of the sensing probe remains constant. As shown in Fig. 2(a), the preset distance is automatically maintained during the lateral movement of the sensing head, after initiating OPD-regulation mode. Hence, the fiber tip moves vertically to maintain the relative distance from the sample at 800 μm, which is set for OPD-regulated OFDR operation.

Figure 2. The schematic and results of the OPD-regulated OFDR system. (a) Schematic of the OPD regulation test on a multi-grooved metal sample with 2-mm intervals and a 0.6-mm dip. (b) The comparison between the off and on operations of the OPD regulation function. OPD, optical-path difference.

Without OPD regulation, when the sensing probe moves laterally over the electroplated sample of PCB, the OFDR sensing probe can bump the sample surface (as the slope of a flat sample is not easily controlled), thus leading to a potential crash. Conversely, with OPD regulation, the distance between the optical-fiber tip’s end and the sample surface is maintained stably, without the probability of a bump. The OPD-regulated OFDR system can be moved vertically with a minimum resolution of 10 μm at once, with a maximum movement range of 78 mm and a response speed of 1 kHz. Once the system is at a specific position to sense the electroplating thickness, the OPD-regulated OFDR operation must be turned off to measure the small change in plating thickness under stable conditions.

3.1. Experimental Setup

Figure 3 shows a schematic of the in situ sensing of phase-based distance for uniformized copper-plating thickness. The electroplating method is implemented with 1.5 L of electrolyte solution composed of 0.25 M H₂SO₄ and 0.5 M CuSO₄, and two Pt mesh electrodes. An electroless-plated copper-clad laminate (CCL) sheet with a thickness of 3 μm, a length of 220 mm, and a width of 180 mm is prepared as a PCB substrate. Power supplies (PARSTAT2273; Princeton Applied Research, TN, USA) A and B are used to apply a constant potential between the electrodes and the CCL sheet. Two OPD-regulated OFDR sensing probes are used to measure the copper-plating thickness in real time. Since the glass fiber of the sensing probe is an insulating and chemically resistant material, the plating thickness can be measured in an electrolytic solution during copper plating. Each sensing probe is moved to the measurement position using the OPD regulation function for safe lateral movement, and positioned at an optimal distance. Next, the OPD-regulation function is turned off, and the plating thickness measurement begins. From the phase variation δφ(z₀) of the interference signal, the phase-based distance variation δd can be monitored in real time. Assuming that the refractive index of the electrolytic solution does not change during the electroplating, the measured δd corresponds to the plating thickness. For the OPD calculation, the refractive index of the electrolyte solution is estimated at 1.332, based on the Arago–Biot approach. From Eq. (2), the measurement range of plating thickness without the 2π ambiguity is predicted as approximately 490 nm. Due to the slow and continuous increment of the plating thickness, it can be measured over 2mm, using the unwrapping method [25–27]. In this experiment, the copper-plating thickness is measured up to 2,325.3 nm by utilizing the unwrapping function.

Figure 3. Schematic setup of copper plating system with two power supplies—A and B; CCL, copper-clad laminate.

To check the stability of this sensing system, the phase-based distance is measured under zero-potential conditions for 30 min. Figure 4(a) shows the result of in situ sensing of the phase-based distance for copper plating, with the distance-sensing variation shown through the probability distribution by measuring the phase-based distance 1,000 times at 800 μm. Figure 4(b) shows the standard deviation of phase-based distance measurements based on the number of averaged data. As the number of averaged data increases, the standard deviation of the distance decreases, to 3.66 nm for an average of 100 data. This result indicates that the stability of the phase-based distance measurement is 3.66 nm, and that the averaging method can reduce the phase error of the measurements efficiently.

Figure 4. The characteristics of phase-based distance stability (a) the probability distribution of phase-based distance of 1,000 measurements at a distance of 800 μm, and (b) the standard deviations according to the average number.

3.2. Experimental Results for Phase-based Distance

To measure the plating thickness in the electroplating process of a PCB, the growth of copper thickness must depend on the voltage applied to the Pt electrodes. Two OPD-regulated OFDR sensing probes are located opposite each other, 80 mm from the center of a CCL sheet with a length of 220 mm. To verify the plating thickness using cross-sectional SEM imaging after electroplating, an insulating tape that prevents further electroplating is attached next to each sensing point.

For the two experiments with different values of the output voltage of power supply A (V_a), V_a is to 0.3 V and 0.5 V respectively, while the output voltage of power supply B (V_b) is set to 0 V. The electroplating is performed for 30 min under each voltage V_a. The plating thickness is measured in situ during electroplating with two OFDR sensing probes, A and B. After electroplating, the insulating tapes are removed, and the plating thickness of each sample is measured using cross-sectional SEM imaging. In these SEM images obtained at the edge of the taping area, the height of each step indicates the plating thickness.

Figures 5(a) and 5(c) show the in situ plating-thickness measurements using the two OFDR sensing probes, when V_a is 0.3 V [Fig. 5(a)] and 0.5 V [Fig. 5(c)]. V_b is set to 0 V. From these results, we confirm that the plating thickness increases continuously during electroplating, and the thickness deviation of both sensing points A and B is related to the voltage difference between the power supplies A and B. The deviation in the plating thickness is proportional to the magnitude of differential current density, which could also be predicted from the diffusion principle of electroplating [28]. Figure 5(a) indicates that the measured value of the plating thickness obtained using OFDR sensing probe A is 1,510.33 nm, and that obtained using probe B is 331.3 nm. Figure 5(b) shows the SEM image of the copper plating with V_a of 0.3 V. The SEM cross-sectional images are recorded at the taping points on the electroplated CCL sample sheet next to probes A and B. From the SEM images, the plating thickness at the sensing point of probe A is 1,540 nm, and that at the sensing point of probe B is 332.3 nm. Compared to the OFDR-sensing measurements, the difference between the results from OFDR and SEM imaging was 30 nm (1.9%) and 1.0 nm (0.3%) at sensing points A and B respectively.

Figure 5. Results of the in situ plating-thickness measurement using the two optical-path difference (OFDR) probes. (a) Plating thickness is measured by sensing probes A and B when V_a is 0.3 V and V_b is 0 V. (b) Cross-sectional image of plating thickness is measured via SEM when V_a is 0.3 V and V_b is 0 V. (c) Plating thickness is measured by sensing probes A and B when V_a is 0.5 V and V_b is 0 V. (d) Cross-sectional image of plating thickness is measured via SEM when V_a is 0.5 V and V_b is 0 V. SEM, scanning electron microscopy; V_a, the output voltage of power supply A; V_b, the output voltage of power supply B.

Similarly, as shown in Fig. 5(c), under V_a of 0.5 V and V_b of 0 V, the plating thicknesses at the sensing probes A and B are measured as 2,325.5 nm and 707.2 nm after 30 min of electroplating. The SEM images at the same points [Fig. 5(d)] demonstrate that the plating thicknesses are 2,310 nm and 673.2 nm at sensing points A and B respectively. The difference between the OFDR and SEM-imaging results is 15.5 nm (0.6%) and 34.0 nm (5.1%) at sensing points A and B respectively. There are some possible explanations for the relatively large error in the results at sensing point B: First, the fixed holder of probe B could have loosened gradually during electroplating. Second, the sample could have been damaged during the cold mount process, a preprocessing method for obtaining SEM images. Third, a human error might have occurred when setting the scale bar for analyzing SEM images.

3.3. Experimental Results for Uniformized Copper-plating Thickness

The second experiment involves uniformizing the copper-plating thickness by monitoring two different positions in real time using OFDR sensing probes A and B, and controlling via proportional voltage tuning of the two power supplies. In this experiment, the voltage condition of power supply A is initially set to a fixed value in the potential mode. Then power supply B is simultaneously used to uniformize the plating thickness, by turning it on or off based on the in situ sensing of the thicknesses at the two positions. The procedure is as follows: The plating thicknesses from the two OFDR-sensing probes A and B are measured in real time. If the plating thickness measured by probe A is thicker than that measured by probe B, then power supply B is automatically turned on. If the plating thickness becomes uniform, i.e. the measurements by probes A and B are equal, then V_b is again set to 0 V.

Figure 6 shows the results of enforcing plating-thickness uniformity during electroplating for 15 min with two OFDR-sensing probes on site. Initially V_a is set to 0.1 V and V_b is set to 0 V. As time passes the copper thickness at the sensing point of probe A becomes proportionally thicker than that at the sensing point of probe B, based on the electroplating time. After 7.5 min the thickness uniformization mode is turned on, and thus V_b increases until the copper thickness at probe B reaches that at probe A. After 4 additional minutes, once the thickness at the two points has almost reached the same value, the thickness uniformization is turned off. Then the plating-thickness deviation between the sensing points of probes A and B increases again, owing to the asymmetric electric field in the electrolytic solution under the initial voltage conditions.

Figure 6. Demonstration of a uniformized copper plating thickness by controlling two power supplies A and B based on in-situ phase-based distances obtained via sensing probes A and B.

In this study, we have proposed an OPD-regulated OFDR sensor system to measure plating thickness in real time, and to implement uniform plating thickness across multiple positions. To avoid a collision between the OFDR-sensing probe and the CCL-sheet sample’s surface, and to regulate the optical distance for high-SNR phase-based distance sensing, the OFDR sensor was combined with feedback motion control via a linear PM. A high response frequency of 1 kHz and stroke length of 30 mm enabled the proposed sensing probe to move without collision, even on a centimeter-scale surface structure. During the copper-plating-thickness measurement via two OFDR sensing probes, the in situ plating-thickness measurement speed was 100 kHz. The proposed OFDR sensor could measure a minimum thickness of 3.66 nm by averaging 100 data points. In this experiment we measured copper-plating thicknesses up to 2,325.3 nm.

Our OFDR sensor is based on a CP interferometer and a flat sensing tip; thus, the SNR is relatively low, compared to that of a conventional swept-source OCT. However, to improve the SNR a balanced detector can be used to suppress the DC envelope of an interferogram, by using a small portion of laser output through a fiber coupler. This method allows the digitizer to use the full-scale range (positive and negative voltages), thereby increasing the voltage resolution. Additionally, since most samples in our application will be metals and the sensing beam is divergent, the optical-power limit for the sample is not considered. When additional sensing probes or a higher SNR is needed, the output power of the laser can be amplified using a boosted optical amplifier [29].

Monitoring the phase-based distances at different sample positions in real time during electroplating makes it possible to successfully measure the plating thickness in situ, and to uniformize the plating thickness across various locations of the PCB substrate. Since our system was based on a CP interferometer without separate reference-beam paths, the number of sensing points could easily be increased by a factor of two. Therefore, if multiple sensing points and electrodes are used in practical applications, ultrasophisticated and advanced electroplating control will be possible.

The authors declare no conflicts of interest.

Data underlying the results presented in this paper are not publicly available at the time of publication, but may be obtained from the authors upon reasonable request.

2-Year Research Grant of Pusan National University.