Polyimide Metallization Using Nickel Nano-Film as both Catalyst and Barrier Layer of Copper Electroless Deposition

Polyimide films were purchased from a commercial polyimide sheet (DuPont, Kapton HN, 50 μm thick). The process for chemical modification of the PI surface is illustrated in Fig. 1. In step (a), the PI film was carefully cleaned in an ethanol solution using an ultrasonic tool for 5 min and then rinsed using deionized water (DI water, 18 MΩ cm). In step (b), the cleaned PI film was immersed into a 5M KOH solution at 50°C for several period of time to hydrolyze the film surface. The details of the alkali-induced modification of the PI surface have been reported in many articles.^44,57,58 The KOH-induced hydrolysis of PI leads to the formation of potassium salts with carboxylic acids via cleavage of the imide ring. The resulting carboxyl groups serve as cation exchangeable groups. In step (c), the surface-modified PI film was immersed into a 50 mM NiSO₄ solution at 25°C for 5 min to incorporate Ni²⁺ ions into the film through an ion-exchange reaction with K⁺ ions. In step (d), the Ni²⁺-doped PI film was immersed into a metallic ion-containing 0.5 M dimethylamine borane (DMAB) solution at 30°C for 20 min to reduce the doped Ni ions. In step (e), electroless copper deposition was carried out by means of the catalysis of various Ni-nFs.

Zoom In Zoom Out Reset image size

The extent of PI hydrolysis by KOH solution was characterized using a Fourier transform infrared spectrometer (FTIR) (Varian 640-IR) with an attenuated total reflection (ATR) attachment. The element analyses of the chemically modified PI film were measured using an X-ray photoelectron spectrometer (XPS) (Ulvac-PHI, PHI 5000) with an Al Kα X-ray source (hυ = 1486.6 eV). The depth profiles of K⁺, Ni²⁺ ions and Ni-nFs in and on the modified PI film were obtained by argon ion bombardment in the spectrometer. The etching rate of the argon ion bombardment was around 24.1 nm min⁻¹. The emission intensity was then converted to an atomic percentage (%).

The electroless copper deposition solution was composed of 0.04M CuSO₄, 0.051M C₆H₅Na₃O₇, 3.8mM NiSO₄, 0.17M NaH₂PO₂, 0.485M H₃BO₃ and 0.5 ppm thiourea. Its pH values was 7 and temperature was 40°C.

The surface morphologies and cross-sections of the samples were observed using field emission scanning electron microscope (FE-SEM, JEOL, JSM-7401F) and transmission electron microscope (TEM, JEOL JEM-1400 Electron Microscope), respectively. For cross-sectional TEM observation, the samples were sectioned into ca. 100-nm-thick slices with the conventional microtome technique using a diamond knife.

To form a layer of PAA on the PI, the PI was immersed in a KOH solution. Figure 3a shows the FTIR patterns of a pristine PI film. Several typical absorption peaks were observed at 1775 cm⁻¹ (imide C=O, asymmetric stretch), 1720 cm⁻¹ (imide C=O, symmetric stretch), 1500 cm⁻¹ (benzene ring, C=C stretch), 1370 cm⁻¹ (imide C–N–C, ring stretch), and 1235 cm⁻¹ (aromatic ether, Ar–O–Ar stretch).^63,66–68 The FTIR pattern shown in Fig. 3a is in agreement with the result shown in Fig. 4a, indicating that only three elements, carbon, oxygen and nitrogen, were revealed by XPS.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

After the PI was immersed in the KOH solution for surface hydrolysis reaction, the absorbance peaks at 1775 cm⁻¹, 1720 cm⁻¹, and 1370 cm⁻¹ disappeared, but the bands at 1500 cm⁻¹ and 1235 cm⁻¹ remained. On the other hand, several new absorbance peaks appeared at 1640 cm⁻¹ (amide, N–H stretch), 1600 cm⁻¹ (carboxylate salt, asymmetric stretch), 1545 cm⁻¹ (amide, N–H bend), 1410 cm⁻¹ (carboxylate salt, symmetric stretch), and 1220 cm⁻¹ (carboxylic acid, C–O), as confirmed by the spectrum shown in Fig. 3b. This result indicates that a layer of poly(amic acid) (PAA) was formed due to immersion in the KOH solution.^63,66–68 Of course, the signals of K⁺ ion was also obtained by XPS, as shown in Fig. 4b, which means that the imide ring on the PI surface was opened and R-COO⁻K⁺ was formed. Following the ion exchange reaction of K⁺ ion with Ni²⁺ ion, the FTIR pattern shown in Fig. 3c is the same as Fig. 3b. However, its XPS pattern significantly shows that the K⁺ signal disappears and Ni²⁺ signal appears, as shown in Fig. 4c, indicating that the doped K⁺ ions in the PAA layer were completely replaced by Ni²⁺ ions.

After the Ni²⁺-doped PI was immersed in the DMAB solution, the doped Ni²⁺ ions in the PAA were chemically reduced by DMAB to form Ni atoms. Hence, the signal intensity of Fig. 3d becomes weak because IR is difficult to tunnel through the Ni-nF. In the meantime, a chemical shift of nickel element occurred, as shown in Fig. 4d, implying that they are not nickel ions but nickel atoms. The performance of the ion exchange between the K⁺ and Ni²⁺ was confirmed by the depth profile analyses of XPS, as shown in Fig. 5. Obviously, element K [see curve (d)] completely disappeared instead of Ni element [see curve (e)]. In addition, the curve (e) shows that the distribution of Ni element is uniform in the PAA.

Zoom In Zoom Out Reset image size

To further confirm this, XPS patterns of nickel element with high resolution were carried out and shown in Fig. 6. Obviously, no nickel metal was detected but nickel ion^69,70 only before reduction reaction using DMAB, as shown in Fig. 6a. After reduction treatment in the DMAB solution, metallic nickel was detected combined with nickel ion, as shown in Fig. 6b. The results show that the doped nickel ions in the PAA were indeed chemically reduced to nickel atoms.^70–72 As to the nickel ions shown in Fig. 6b, two possible reasons can explain it. The first possible reason is nickel oxidation in air with moisture after reduction reaction; the second possible reason may be attributed to that some of the doped nickel ions were not chemically reduced by DMAB.^69–72 According to the relative signal intensities of curves (a) and (b), the nickel ions of curve (b) should be attributed to oxidation reaction with moisture because the doped nickel ions are buried in the PAA, which results in weak intensities, as shown in Fig. 6a.

Zoom In Zoom Out Reset image size

A lot of papers have reported that the barrier layer, such as Ni-nF, is very important for PI metallization, because copper atoms easily diffuse into both PAA and PI film.^1,73,74 Hence, before copper electroplating onto the PI film, a continuous and dense barrier film has to deposit onto the PI film first. Herein, we developed an approach that can form a layer of continuous and dense Ni-nF onto the PAA. The Ni-nF also possess catalytic activity for electroless copper deposition in order to enhance the conductivity of the metallized PI for the followed copper electroplating.

Figure 7 shows that the rapid electron release of DMAB in the presence of Ag⁺, Cu²⁺ and Pd²⁺ ions (see the LSV analyses shown in Fig. S1) to the nickel ions on the PAA surface, it can facilitate the upward diffusion rate of the doped nickel ions. This effect of rapid electron release of DMAB can be carried out by metallic ion addition (i.e., Ag⁺, Cu²⁺ and Pd²⁺) in the DMAB solution (see Fig. S1) to form a layer of dense Ni-nF on the PAA. Based on the grain density, Fig. 7A shows that the optimal concentration of AgNO₃ is 1.06 ppm shown in Fig. 7A(e). Figure 7B shows that the optimal concentration of CuSO₄ is 2.5 ppm shown in Fig. 7B(e). And Fig. 7C shows that the optimal concentration of PdCl₂ is 0.84 ppm shown in Fig. 7C(e). These results are in agreement in the results shown in Fig. 2, which explain that the trend of electroless copper deposition rates with the salt concentration agree with the SEM morphologies, as shown in Fig. 7.

Zoom In Zoom Out Reset image size

To further confirm that Ag, Cu and Pd ions can enhance DMAB oxidation and induce the doped Ni ions in the PAA prefer to aggregate on the PAA surface because of the rapid reduction of the surface Ni ions and correspondingly leading to a steep concentration gradient of Ni ion, HR-TEM was employed to image the PI cross-sections. Figure 8 shows the typical HR-TEM images, demonstrating that a lot of Ni nanoparticles were present in the PAA, as shown in Fig. 8a, because there was no metallic ion in the DMAB solution. The others, such as shown in Figs. 8b–8d, almost no Ni nanoparticle was present in the PAA but all Ni ions were reduced to Ni and aggregate to form a film on the PAA. The best performance appears in Fig. 8b, which means that Ag⁺ ion can effectively induce the doped Ni ions in the PAA to rapidly diffuse toward the PAA surface.

Zoom In Zoom Out Reset image size

Figure 2 further demonstrates that the reduced nickel ions do exhibit a steep concentration gradient in the PAA. Interestingly, two waveforms, indicated by the arrows, can be obviously obtained in the presence of Ag⁺ (see Fig. 2c), Cu²⁺ (see Fig. 2b) and Pd²⁺ (see Fig. 2d) compared with the case without those metallic ions (see Fig. 2a). The distance between the two waveforms (i.e., two peaks) shows that curve (b) ≅ curve (c) < curve (d) ≪ curve (a). The results indicate that Ag and Cu ions can rapid induce the doped Ni ion to diffuse toward the PAA surface, so that the aggregated Ni atoms distribution has a shorter distance compared with the case without metallic ion induction (i.e., curve (a)). Accordingly, the residual Ni ion of curve (a) was higher than the other curves, as shown in Fig. 2a.

A mechanism that can explain the effect of those salt addition in the DMAB solution is illustrated in Fig. 9 according to the results and Figs. 2 and 8. The scheme (A) shown in Fig. 9 describes the process of the Ni-nF formation without the salt addition. Finally, the Ni atoms will aggregate together to form nanoparticle on and in the PAA. On the other hand, the scheme (B) shown in Fig. 9 indicates the process of the Ni-nF formation with the salt addition. Finally, the Ni atoms will aggregate together to form the Ni-nF on the PAA because of the rapid electron release of DMAB catalyzed by those metallic ions. Therefore, there are almost no Ni nanoparticles but dense Ni-nF on the PAA, as illustrated in Fig. 9g. Of course, those metallic ions may be also chemically reduced by DMAB and incorporated into the Ni-nF. The issue was verified by secondary ion mass spectrometer (SIMS) analyses by adding 1.5 and 3.0 ppm salts in the DMAB solutions. As a result, Fig. S3 and S4 show that both cases cannot detect Ag, Cu and Pd elements, respectively. The results indicate that those metallic ions should only play the role of catalyst to facilitate DMAB to rapidly release electrons to reduce the doped Ni ions.

Zoom In Zoom Out Reset image size

To further confirm that DMAB is able to diffuse into the PAA and reduce the Ni ions in the PAA to be the Ni nanoparticles, looking like the scheme illustrated in Fig. 9d, a comparative experiment was carried out and shown in Fig. S5. Significantly, there were a lot of Ni nanoparticles in the PAA, as confirmed in Fig. S5(a, b), because the Cu²⁺ ion concentration was higher than 3 ppm. It may cause rapid DMAB oxidation and Cu²⁺ ion reduction. Hence, its final mode is similar to that of Fig. 9d, which means that DMAB is indeed able to diffuse into the PAA. On the other hand, the reducer was replaced by H₂ gas, the mechanism would be similar to the case illustrated in Fig. 9g. There was no Ni nanoparticle in the PAA, instead, a Ni-nF was formed on the PAA surface, because H₂ is able to rapidly release electron to the Ni ion on the PAA surface.

Figure 10 summarize the SEM images of after the electroless copper deposition and the corresponding salt concentration added in the DMAB solutions. The salt all has an optimal concentration for the electroless copper deposition rate. These optimal concentrations are exactly in agreement with the SEM morphologies shown in Fig. 7. Since the Ni-nF is a catalyst of the electroless copper deposition, its morphology certainly would influence its catalytic activity and corresponding deposition rate. Many small copper particles were observed in Figs. 10A, 10C. It can be attributed to that Ag and Pd are better catalysts than Cu for the electroless copper deposition. Therefore, these copper particles may be caused by a trace of randomly distributed Ag and Pd atoms. In fact, Fig. 11 confirms that the best Ni-nF catalyst is obtained from the Ag⁺ addition in DMAB, because its copper deposition rate was two times higher than those caused by Cu²⁺ and Pd²⁺ addition. Whatever, three metallic ions exhibit an optimal concentration that can lead to the maximum electroless copper deposition rate, which is in agreement with the results shown in Fig. 7, Fig. 10 and Fig. 11.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size