Patterning micro and mini-features on thermoplastic substrates is a primary concern, especially in developing microfluidic devices mainly used in life sciences for analytical, diagnostic, or therapeutic purposes. The microfluidic devices are composed of microchannels and reservoirs with at least one characteristic dimension (typically the height or the width of features), typically less than 1 mm. Soft lithography [1], micro-milling [2], or hot embossing [3] are the methods used for prototyping or low-volume production of microfluidic devices for testing purposes. On the other hand, injection molding [4, 5] is the standard method for high-volume manufacturing for commercial purposes.
In recent years, ultrasonic embossing has appeared as a promising alternative, especially for medium- to high-volume production of plastic microfluidics, since it reduces the cycle time for patterning the microfeatures on a single microfluidic device to the order of 101 s [6]. Besides, the process can be carried out on an ultrasonic welding machine, which is typically less costly than an injection molding machine [7].
In ultrasonic embossing, a horn vibrating at 20–30 kHz frequencies compresses the thermoplastic substrate against a mold. Figure 1 illustrates the process. In the first stage of the process, the horn presses the substrate against the mold without vibration. The horn starts vibrating in the next stage while pressing on the substrate, generating heat due to friction at the interface surfaces. The heat increases the temperature in the process-affected zone above the glass transition temperature of the substrate material [7], and the softened substrate is deformed. In the following stage, the vibration ceases, and the substrate cools down. Following this hold period, the horn is removed, and the substrate is released.
Comprehensive studies have been carried out to explain the underlying physics of the process [7–9], and the process has been successfully applied to the manufacturing of microfluidic channels [7, 10], optical diffusers [11], and even circuit boards [12, 13].
Several other studies have investigated the effect of the process parameters on replication quality. Qi et al. [14] examined the effect of process parameters on the uniformity and depth of the replicated microstructures. They utilized silicon molds to replicate a series of micro-grooves on polymethyl methacrylate (PMMA) substrates. They stated that the vibration time has the most significant effect on the replication depth. On the other hand, they observed that increasing the pressure improved the replication depth. However, they reported that excessively increasing the pressure reduced replication depth.
Mekaru et al. [15] tested the replication of electroformed nickel patterns on polycarbonate (PC) sheets. They qualitatively assessed the replication quality using an optical microscope to examine the replicated pattern. They stated that the vibration time is the most influential parameter. They also claimed that excessive pressure might suppress the effect of ultrasonic vibration and reduce replicability.
Zhu et al. [16] examined the replication of laser-ablated features on aluminum molds on polyethylene terephthalate (PET) films. They compared the depth of replicated features with the mold to determine the replication rate. They reported that the replication rate increased with increasing pressure but decreased after a specific value. Similarly, they reported that increasing the vibration time improved the replication rate but dropped when it was excessive. They also stated that a longer holding time improved the replication rate.
Liu et al. [11] tested the replication of microfeatures on PMMA with a sawtooth profile machined on mild steel. They used the deviation between replicated feature height at the center and the side of the substrate as the figure of merit. As opposed to the findings in Ref. [15], the authors concluded that higher pressure would result in better replicability. Similarly, they argued that increasing the vibration time would result in better replication. However, they also stated that over time might cause over-melting of the substrate, and replicability might decrease. Similar to the other works, they concluded that a higher hold time would improve replicability.
In another study, Liu et al. [6] examined the effect of process parameters (vibration time, pressure, hold time) by replicating pyramidal features on various polymer substrates, namely PMMA, polyvinyl chloride (PVC), polyethylene (PE), and polypropylene (PP). The depths of the replicated pyramids were used to characterize the performance. They observed that amorphous polymers (PMMA, PVC) exhibited better replicability than semi-crystalline polymers. Similar to Ref. [11], the authors concluded that increasing vibration time, pressure, and hold time, improved the replicability.
Fang-Yu et al. [17] investigated the replicability of PP and PMMA by using electroformed nickel molds with groove-shaped micropatterns. They utilized the height ratio, defined as the ratio of the replicated pattern height to that of the mold, as the figure of merit. They stated that the vibration time was the most influential factor. They also reported that increasing the pressure resulted in a better replication, similar to Ref. [6, 11].
The mold and substrate can be preheated in a variant of the process to enhance the replication quality. Qi et al. [18] investigated the process parameters in thermal-assisted ultrasonic embossing and local thermal-assisted ultrasonic embossing, where the substrate is preheated locally at the deformation zone. In the process, a PMMA substrate was locally preheated by placing it on a heated mold. The average and standard deviation of the depths of a series of replicated features were utilized as indicators of replication quality. They reported the vibration amplitude, the vibration time, and hot plate temperatures as the most effective factors on replication uniformity while increasing the hold time and pressure was observed to improve the replication quality.
As a different approach, Luo et al. [19] examined the effect of the process parameters in the embossing of microstructures etched on silicon molds simultaneously on both the upper and lower surfaces of PMMA substrates. They utilized a hot plate at the mold base as an additional heat source to assist the process. The feature depths on both sides of the substrate were measured to assess the effect of the process parameters. They concluded that the main parameters affecting the replication quality are the vibration amplitude and the thermal-assisted temperature. They also stated that the ultrasonic force is the most effective on replication uniformity.
The abovementioned studies commonly concluded that vibration time is the most effective parameter in ultrasonic embossing. It was also stated commonly that the delay time is not an essential factor in the process. As opposed to delay time, increasing the hold time was observed to improve the replication quality. On the other hand, there was no common ground for the effect of the pressure applied on the substrate by the horn. Some of the studies claim that increasing pressure improves replicability [6, 11, 17], while it is stated that excessive pressure reduces the replication quality [15, 16]. Table 1 summarizes the literature investigating the replication quality in ultrasonic embossing. The difference between the conclusions in different studies might be caused by the difference in the feature dimensions tested or the range of process parameters investigated. On the other hand, it might as well be arousing because a single dimension – depth – of the replicated features was used as the figure of merit. However, the aim is to replicate not only the depth but the entire profile of the features. Therefore, we claim that replicated feature’s profile and its fidelity to the targeted profile should be examined for a precise assessment.
In this paper, we assess the replication quality of ultrasonic embossing by using the cross-correlation function. Cross-correlation is a commonly used tool to determine the similarity between two time series in signal processing. In manufacturing, the utilization of cross-correlation has been presented for tool wear monitoring purposes [20–22]. The method was also utilized for the assessment of the replication of rough surface topographies generated by 3D printing in different polymers by injection molding and polymer casting [23]. Although cross-correlation between the mold and the replication profiles was justified for assessment of replication of generated rough surfaces, an improved method that is more sensitive to local slopes is required in assessing the replication quality of microchannels as the side walls of the microchannels are often designed to be vertical with ideally infinite slope. As an improved method, we proposed to utilize the derivatives of the fabricated and target profiles in computing the cross-correlation. For testing purposes, we fabricated a brass mold including three protrusions in the form of straight lines of varying widths by machining. The protrusions were replicated on PMMA by ultrasonic embossing to form straight channels. The vibration time and the pressure were altered to observe their effects on replication quality. Replication quality was assessed by utilizing both the cross-correlation between the profiles and the cross-correlation between the gradients of the profiles to prove the sensitivity of the proposed method. The effect of the process parameters on replication quality was investigated by using the proposed method.