The effect of surface morphology on the peel performance of UV-induced adhesion-reducing adhesives

In this paper, end-hydroxy fumaryl chloride-diol copolyesters (EHFDCP) with different double bond contents were prepared by the reaction of fumaryl chloride and diols. The molecular weight of the target hydroxy polyesters was controlled to be essentially the same, EHFDCP prepared from 1,8-octanediol, 1,5-pentanediol and ethylene glycol named EHFDCP-1, EHFDCP-2 and EHFDCP-3, respectively. The UV-induced adhesion-reducing adhesives (ARA) were prepared with EHFDCP, isophorone diisocyanate (IPDI), chain extender and photoinitiator. The ARA-1, ARA-2 and ARA-3 were produced by EHFDCP-1, EHFDCP-2 and EHFDCP-3, respectively. After UV curing, all the surfaces of ARA-1, ARA-2, and ARA-3 had a high number of concave and convex areas, which helped to reduce the contact area between the adhesive and the substrate surface. The surface roughness of ARA-2 is the highest and the adhesion reduction effect is the most significant. The higher surface roughness of ARA-2 came from moderate double bond content and crosslinking density. With the increasing photoinitiator content, the 180° peel strength after UV curing decreased. The 180° peel strength of ARA-2 was reduced to 0.16 N/25 mm at 4 wt% of photoinitiator content, and it also had a high initial 180° peel strength of 18.55 N/25 mm due to the absence of small molecule polyfunctional monomers.


Introduction
Semiconductor chips are indispensable basic materials in large-scale integrated circuits and electronic devices [1,2]. The processing of semiconductor chip involves grinding and cutting of wafers, which is thin, brittle, and easy to break and fly apart during processing, so a peelable special tape is needed to bond and fix the wafers [3][4][5]. The tape has a high adhesive strength, which allows it to hold silicon wafers during processing [6]. After processing, the tape loses its adhesive properties and the wafers can be easily picked up [7,8]. Some adhesionreducing adhesives have been reported in recent years, and they can be divided into three types: degradationinduced debonding, and light/heat and UV induced crosslinking debonding [9][10][11][12].
The adhesives of degradation-induced debonding compose of degradable polymers [13]. The adhesive broke down into small molecules, and the peel strength reduced considerably when it was exposed to light or heat [14]. Sugita et al [15] used decomposable multi-acrylate oligomers to prepare an adhesion-reducing adhesive, after post-exposure-bake, the adhesive decomposed into olefin and carboxylic acid derivatives, and debonded from glass panels. Sasaki et al [16] prepared a novel photo-separable adhesive base on poly (olefin sulfone), which showed high adhesive strength on quartz plates. When the adhesive was exposed to UV light and heated to 100°C, the poly (olefin sulfone) depolymerized and released gaseous material resulting in a drastic decreasing in bond strength. However, the degradation of polymers to small molecules usually takes a long time and produces volatile gases. Furthermore, the residues of adhesive are difficult to remove completely.
The adhesives of light/heat induced dual curing debonding are generally prepared from various acrylate polymers, reactive diluents, photoinitiators, and heat curing agents [17][18][19] adhesive decreased with the increase content of the thermal cross-linker KL-1202. However, this curing process has high energy consumption, low efficiency, high cost, and poor debonding ability.
Another candidate strategy is the adhesives with UV-induced crosslinking debonding [21][22][23]. After UV irradiation, the adhesive shrinks greatly in volume through the polymerization of multifunctional oligomers, resulting in a significant reduction in peel strength [24,25]. This method has outstanding advantages such as fast reaction rate, high efficiency, and low energy consumption [26][27][28][29][30]. Hao et al [31] used polyethylene glycolmodified isophorone diisocyanate oligomer (PEG-IPDI) and acrylic acid copolymer to prepare adhesionreducing adhesives. The peel strength significantly reduced and the residual adhesive was very low after UV irradiation due to the formation of an IPN structure. Han et al [32] used 2-isocyanatoethyl methacrylate to modify a hydroxyl-containing acrylate copolymer, and further mixed with a photoinitiator and aziridine crosslinker to prepare the adhesive. The adhesive exhibited good debonding ability after 40 s of UV irradiation and was suitable for wafer cutting and grinding processes. However, the addition of multifunctional oligomers may weaken the cohesive force of the adhesive and reduce the initial peel strength [33][34][35]. In addition, unreacted oligomers can easily transfer to the wafer surface and cause contamination [36].
In this paper, end-hydroxy fumaryl chloride-diol copolyester (EHFDCP) was designed and synthesized. A series of UV-induced adhesion-reducing adhesives were prepared by reacting the hydroxyl group of EHFDCP with the isocyanate group of IPDI, and adding chain extender and photoinitiator. The initial 180°peel strength was high due to the absence of small molecule polyfunctional monomers, and the 180°peel strength after UV irradiation decreased significantly due to the crosslinking of the adhesive to form a concave convex surface structure. This paper reveals the effect of surface morphology on the peel performance of UV-induced adhesionreducing adhesives.

Preparation of EHFDCP
1,8-octanediol (15.18 g), triethylamine (16.16 g) and 4-tert-butylpyrocatechol (0.03 g) were dissolved in 50 mL dried THF. The mixture was added to a 250 mL three-necked flask equipped with a mechanical stirrer, constant pressure funnel, and nitrogen inlet. Under the protection of N 2 , 40 mL THF containing 12.24 g fumaryl chloride was added dropwise through the constant pressure funnel at 5°C, and the reaction was carried out for 1 h. Then the reaction solution was warmed up to 30°C and reacted for 2.5 h.
Triethylamine hydrochloride byproduct was removed by centrifugation, and the product was obtained by precipitation using deionized water, then washed with deionized water several times. Finally, the polymer product was dried under high vacuum at 50°C for at least 8 h to obtain EHFDCP-1 [37].
The synthesis step of EHFDCP-2 and EHFDCP-3 was the same as that of EHFDCP-1 by using 1,5pentanediol and ethylene glycol instead of 1,8-octanediol, respectively. The synthesis route was shown in scheme 1.

Preparation of UV-induced adhesion-reducing adhesive (ARA)
EHFDCP-1 (5 g), IPDI (2.250 g), and dibutyltin dilaurate (0.015 g) were dissolved in 15 mL ethyl acetate. The molar ratio of NCO/OH was fixed at 1.2. The mixture was added to a 100 ml three-necked flask equipped with a Scheme 1. The synthesis route of EHFDCP-1, EHFDCP-2 and EHFDCP-3. mechanical stirrer, reflux condenser, and nitrogen inlet. Under the protection of the nitrogen, the mixture reacted at 60°C for 3 h. Then, the chain extender 1,4-butanediol (0.5 g) was added into the flask, and the reaction continued for 1 h. Finally, 1-hydroxycyclohexyl phenyl ketone (0.233 g) was added to the solution and the mixture was stirred at room temperature for about 1 h to obtain the isocyanate-terminated ARA-1.
In a similar manner, ARA-2 and ARA-3 were prepared from the EHFDCP-2 and EHFDCP-3, respectively. The synthesis route was shown in scheme 2.

Preparation of the ARA tape
The ARA was coated onto the polyolefin film treated by corona with a thickness of 50 μm using wet film applicator and then aged at room temperature for 24 h.

Hydroxyl value
The hydroxyl value was measured according to the method described in ISO 2554-1974.

Fourier transform infrared (FTIR) spectroscopy
The FTIR spectra were measured on a Nicolet iS50 (Nicolet, USA) FT-IR Spectrometer by coating samples on KBr chips. The wavenumber range was 400-4000 cm −1 , and the resolution was 2 cm −1 .

1 H NMR spectroscopy
The 1 H NMR spectra were recorded using a VNMRS 600 spectrometer (Agilent, USA). All samples were analyzed using CDCl 3 as solvent and tetramethylsilane (TMS) as internal reference.

Gel permeation chromatography
Average molecular weight and molecular weight distribution were determined by a GPC instrument (Waters 1525, USA). Polystyrene standards were used for calibration and THF was used as carrier solvent.

180°peel strength
The silicon wafers cutted into 25 mm wide were glued to the stainless-steel plate, the tape was attached to the wafers and a 2 kg rubber roller was rolled over them twice. The samples were allowed to place at room Scheme 2. The synthesis route of the prepolymer. temperature for 24 h to measure the 180°peel strength of before and after UV irradiation. The sample was cured by using a 1KW UV lamp (the wavelength of UV lamp is 365 nm) to irradiate 60 s. The 180°peel strength of the tape was measured using a universal testing machine CMT4304 (Shenzhen Sans, China) according to ISO 29862: 2007. The peeling speed was 300 mm min −1 .

SEM
The surface morphology of the samples was observed by a field-emission scanning electron microscope Gemini 500 (ZEISS, Germany). All samples were sprayed with a thin layer of gold before SEM observation.
2.5.7. Laser confocal 3D measurement microscope 3D surface morphology observation of the samples was carried out with a laser confocal 3D measurement microscope VK-X250 (KEYENCE, Japan). The resulted pictures were further analyzed and processed with MultiFileAnalyzer software. The surface roughness parameters including the arithmetic average height (Sa), root mean square height (Sq), maximum peak height (Sp), maximum valley height (Sv) and the maximum height (Sz) were obtained.

Results and discussion
3.1. Synthesis and characterization of EHFDCP Diols containing methylene groups of varying lengths (1,8-octanediol, 1,5-pentanediol, and ethylene glycol) were used to react with fumaryl chloride for synthesizing polyesters with hydroxyl groups as end groups. The molecular weight of the target hydroxy polyesters was controlled to be essentially the same, and the content of carbon-carbon double bond in the main chain gradually increased from EHFDCP-1 to EHFDCP-3. The molar ratios of diols to chloride in the feedstock were listed in table 1. As shown in figure 1, the number-average molecular weights (M n ) of EHFDCP-1, EHFDCP-2 and EHFDCP-3 were 1282, 1482 and 1404 g mol −1 ,  respectively, and the hydroxyl value of EHFDCP-1, EHFDCP-2 and EHFDCP-3 were 64.83, 69.91 and 74.05 mg KOH/g, respectively.
1 H NMR spectra of EHFDCP-1, EHFDCP-2 and EHFDCP-3 were shown in figure 2. In the spectrum of EHFDCP-1 ( figure 2(a)), the peak at 6.84 ppm was attributed to the protons of -CH=CH-, the peak at 3.63 ppm was assigned to the protons of C-OH, and the peaks at 4.18 and 1.33-1.67 ppm corresponded to the protons of -CH 2 -CH 2 -CH 2 -CH 2 -CH 2 -CH 2 -CH 2 -CH 2 -. Similar results were presented in the spectra of EHFDCP-2 and EHFDCP-3 [39]. As shown in figure 2(b), the peak at 6.82 ppm was assigned to the hydrogen protons of -CH=CH-, the proton peaks of the long chain -CH 2 -CH 2 -CH 2 -CH 2 -CH 2were found at 4.19 and 1.44-1.72 ppm, the peak of C-OH group was seen at 3.63 ppm. In the spectrum of EHFDCP-3 (figure 2(c)), the proton peak of −CH=CH-, C-OH, and -CH 2 -CH 2was at 6.89 ppm, 3.89 ppm, and 4.34-4.66 ppm, respectively. The solvent peak of the deuterated chloroform (CDCl 3 ) was at 7.26 ppm.  The FT-IR spectra of EHFDCP-1, EHFDCP-2, and EHFDCP-3 were shown in figure 3. The stretching vibration peak of C=C appeared at 1644 cm −1 , and became weaker as the increasing length of -CH 2 -. The absorption peak at 3446 cm −1 was attributed to the stretching vibration of hydroxy, the peak at 2858-2929 cm −1 represented the stretching vibration of C-H, and the absorption peaks at 1720 cm −1 , 1160 cm −1 were ascribed to C=O stretch and C-O stretch patterns, which confirmed the formation of ester bond in synthesis of EHFDCP [40].

Adhesion reduction and surface morphology
The 180°peel strength of initial and after UV irradiation of ARA-1, ARA-2 and ARA-3 were measured and shown in figure 4. Before UV irradiation, the 180°peel strength of ARA-1, ARA-2 and ARA-3 were 16.5, 18.2 and 17.4 N/25 mm, respectively. All samples showed high initial 180°peel strength. The reason may be that there were no small molecule multifunctional monomers in the binder, and cohesion hardly reduced by plasticizing. In addition, the hydrogen bonding in carbamate (NHCOO), carbonyl group (C=O) and silicon wafer surface might further strengthen the initial bond strength [41,42]. After UV irradiation, the 180°peel  strength of ARA-1, ARA-2 and ARA-3 decreased to 3.68, 0.46, 2.14 N/25 mm, respectively. The decrease of the 180°peel strength may be due to the adhesive shrank during the UV curing process. The most significant adhesion-reducing effect was obtained for ARA-2. It is likely that a special morphology formed by ARA-2 of moderate double bond content is benefit to reduce the adhesion during UV curing process. Figure 5 shows the surface morphology of ARA-1 to ARA-3 after UV irradiation. The surfaces of ARA-1, ARA-2 and ARA-3 all showed a large number of concave and convex areas, which were due to the volume shrinkage caused by crosslinking of the adhesion-reducing adhesive under UV irradiation. The generation of the bumpy structure reduced the contact area between the adhesive and the surface of the substrate, resulting in reduced cohesion. In this study, ARA-2 showed the best adhesion reduction effect because it possessed the most obvious concave and convex structure and the maximum number of concave and convex areas in the same unit area compared with the other two samples.
Laser confocal 3D measurement microscope was used to observe the surface and 3D height images of ARA-1, ARA-2 and ARA-3 before and after UV curing, and the results were shown in figure 6. Before UV irradiation, the surface of ARA-1, ARA-2 and ARA-3 was very flat, which could facilitate the wetting and adsorption of ARA on the wafer surface, resulting in a high initial 180°peel strength. After UV curing, at the same scale, all three samples showed concave and convex structures on the surface, which reduced the contact area between the adhesive and the surface of the substrate, resulting in the decrease of 180°peel strength, but compared to ARA-1 and ARA-3, the concave and convex structures of ARA-2 were more obvious. This phenomenon was also confirmed by the 3D height images, where the Z-axis value of ARA-2 (11.94 μm) is much larger than that of ARA-1 (3.85 μm) and ARA-3 (6.77 μm). The surface roughness was usually used to characterize the concave and convex structure [43,44]. The values of a series of roughness parameters, including Sa and Sq, were shown in table 2. It can be clearly seen that the Sa values of ARA-1, ARA-2 and ARA-3 were 0.34, 1.01 and 0.50 μm, respectively, and the Sq values were 0.44, 1.28 and 0.64 μm, respectively. The other three roughness parameters including Sp, Sv and Sz also revealed that the corresponding value of ARA-2 was the highest. Sdr indicates how much the unfolded area of the defined region increases compared with the area of the defined region. The Sdr of    ARA-1, ARA-2 and ARA-3 were 0.06, 1.02, and 0.38, respectively. The above roughness data showed that ARA-2 has the highest surface roughness.
The crosslinking density has great influence on the surface roughness of polymer [45]. Under UV irradiation, the crosslinking of adhesion-reducing adhesive increased the intermolecular forces and caused uneven stresses during UV curing, resulting in volume shrinkage and uneven surface. The Ve value after UV curing of ARA-1, ARA-2 and ARA-3 were 5.55×10 −3 , 13.32×10 −3 and 31.21×10 −3 mol cm −3 , respectively. As the double bond content increased, the crosslinking density of ARA increased. At the proper crosslink density, the surface was the roughest, the concave and convex structure of the surface was more obvious, and the adhesion reduction effect was better. Yong et al investigated the effect of crosslinker content on the surface microstructure of acrylic resin film. The results showed that when the crosslinker content increased from 0 to 0.2 wt%, the surface of the film became rough and presented a bumpy structure, and when the crosslinker content was further increased to 0.4 wt%, the surface of the film did not change much. The surface roughness is maximized under the appropriate crosslinking degree [46]. Figure 7 shows the effect of the amount of photoinitiator on the 180°peel strength of ARA-2 before and after UV irradiation. Before UV irradiation, the initial 180°peel strength did not change much due to the addition of a small amount of photoinitiator, while after UV irradiation, the 180°peel strength decreased with the increase of photoinitiator content, and the 180°peel strength did not change much when the photoinitiator content was higher than 4 wt%. With the increase of photoinitiator concentration, the concentration of free radical increased and the crosslinking reaction accelerated, which resulting in an increase of crosslinking density and a decrease of 180°peel strength. At 4 wt% photoinitiator content, the 180°peel strength was 18.55 N/25 mm before UV irradiation and reduced to 0.16 N/25 mm after UV curing. The 180°peel strength of tape produced by LINTEC Corporation was about 11.38 N/25 mm before UV irradiation and 0.56 N/25 mm after UV irradiation [31]. Figure 8 shows the surface and three-dimensional height images of ARA-2 after UV curing with different photoinitiators contents. The concave and convex microstructure is thought to help reduce the contact area between the adhesive and the substrate, thus achieving the adhesion reduction effect. As shown in figure 8, as the photoinitiator content increased from 1 wt% to 4 wt%, the surface roughness became more obvious and the 180°peel strength decreased from 1.52 N/25 mm to 0.16 N/25 mm after UV irradiation, which was mainly because the increasing photoinitiator content increased the concentration of free radicals and accelerated the crosslinking reaction. When the photoinitiator content was increased to 5 wt%, the surface bumpy structure changed very little and the 180°peel strength after UV curing hardly changed.

Effect of photoinitiator content on surface morphology
The roughness data of ARA-2 with different photoinitiators contents were given in table 3. With the increase of photoinitiator content, Sa increased from 0.88 μm to 1.19 μm, Sdr increased from 0.68 to 1.37, and reached the maximum when the content of photoinitiator was 4 wt%. The other roughness parameters including Sz, Sp, Sq and Sv also followed this trend. When the photoinitiator content increased to 5 wt%, the roughness parameters were roughly the same as that of 4 wt% photoinitiator.

Conclusion
Three kinds of EHFDCP with different double bond contents were synthetized and confirmed by FT-IR and 1 H NMR, and three UV-induced adhesion-reducing adhesives were successfully prepared. All samples had high initial 180°peel strength due to the absence of small molecule multifunctional monomers. After UV curing, the different crosslinking density caused a difference in the surface morphology of the adhesive, which further resulted in the variable adhesion reduction effect. Compared with ARA-1 and ARA-3, the surface concave and convex structure of ARA-2 was the most obvious, which resulted in the best adhesion reduction, and the 180°p eel strength decreased from 18.21 N/25 mm to 0.46 N/25 mm. The 180°peel strength of ARA-2 after UV irradiation was further reduced to 0.16 N/25 mm at 4 wt% photoinitiator content.