1 Introduction

Three-dimensional (3D) integration, which uses through-silicon vias (TSVs) to interconnect multiple active circuit layers, is a central issue of electronics packaging that encompasses different types of technologies. Through-Silicon-Via is widely accepted in industry because it performs high density interconnects with a good electrical performance and a small form factor for interconnecting stacked devices on wafer level [1–3]. Although TSV technology is conceptually simple, there are many issues to resolve for high volume manufacturing because of its complicated manufacture processes [4].

Via filling with void-free is still a major challenge in these processes, albeit a number of approaches have been developed for copper filling, such as PVD, CVD and electrodeposition [2, 5, 6]. Among these approaches, copper electrodeposition is efficient and cost-effective, which is widely approbated in industry. However, TSV copper filling is easily affected by electrodeposited process, including composites of electrolyte with different additives, electric field, copper seed layer as cathode and pretreatment [2, 7–14]. Among these processes, it is always necessary that copper seed layer is repaired by special process because of limitation of fabrication procedure [7, 9, 15]. Meanwhile, pre-wetting as a step of standard operation procedure (SOP) of TSVs is scarcely reported in researches. With increasing high-aspect ratio of via, potential hazards of air trapped in via will be greatly enhanced when chip with TSVs is directly immersed in electrolyte without efficient pretreatment. Although utilizing stirring with ethanol in pre-wetting process to achieve void-free deposition is reported by Moffat and Dow, details of pre-wetting are not illustrated [2, 16–18]. Compared with ethanol as pre-wetting liquid with high speed rotation, vacuuming with pre-wetting liquid demonstrate the ability to pre-wet the high aspect ratio [19, 20]. In our former research, the disruption of copper seed layer has happened when ultrasonics is used as pretreatment. Furthermore, vacuum is a more efficient method to pre-wetting the whole sidewall of via with big high-aspect ratio, particularly those with narrow gates [21]. Meanwhile, it is a disaster that copper oxide exists on the surface of seed layer. According to that copper oxide is insulator, it leads to copper electroplating invalidly as copper oxide exists on the surface of seed layer during copper filling. The copper oxide is cleaned by dilute H2SO4 which is proposed by Lee et al. [22]. But the integration of seed layer is not considered in their research. Hence, it is of very interest to develop a one-step pretreatment wet via before the chip with TSVs is immersed into electroplating bath, especially copper seed layer is oxidized on the surface.

In this paper, we propose a pretreatment to assure copper filling with void-free, even if the surface of copper seed layer is oxidized. And it can integrate the two processes, such as air expel and removing oxide, into one. Basic Electrolyte Solution (BES) which is preferred to be pre-wetting liquid was investigated in the process because of its characteristic. The effect of the pretreatment is verified by experiment of TSV copper filling.

2 Experiment

Wettability measurements were performed on chip with TSVs. TSVs with three different aspect ratios were on the silicon wafer, which was covered by insulation layer SiO2, barrier layer Ti/TiN and seed layer Cu (150 nm). The size of the three TSVs was 20, 30 and 40 μm in diameter and 90, 95, and 100 μm in height, respectively. TSVs distribute periodically on the silicon wafer. It is provided by Shanghai Sinyang Electronics Chemicals Co., Ltd., China.

Contact angle (CA) between wetting agent and Cu seed layer were measured with a 3.5 μL solution droplet at ambient temperature using an optical contact anglemeter (Contact angle measurement, JC2000C1). These measurements were performed at 25 °C.

Thickness measurements were performed to obtain the thickness of copper seed layer and copper oxide film in chip. Three different test techniques were performed for objectively measurement, which included polarization resistance method, X-ray Fluorescence thickness measurement (FISCHERSCOPE XULM, Fischer) and Probe Profilemeter (Dektak XT, Bruker). The polarization resistance method which used the chip with TSVs as the cathode, was performed in Sodium hydroxide solution by Electrochemistry workstation (CHI660D, Shanghai Chenhua Instrument Co. Ltd.).

The statue of copper seed layer is examined before the copper filling experiment. It is impossible to examine the quality of copper seed layer in thorough via on the chip before performing the copper filling in via. Hence, the uniformity of copper seed layer is examined by Probe Profilemeter on the surface of chip. According to the manufacturing process, the uniformity of copper seed layer is more difficultly generated with increasing high-aspect ratio of via [9]. Hence, the quality of copper seed layer in via with 20 μm in diameter and depth in 90 μm is examined by FE-SEM as shown in Fig. 1.

Fig. 1
figure 1

The schematics of experimental device for copper filling and detailed illustration of cathode with chip

Full size image

The composites of BES were 110 g/L Cu(CH3SO3)2, 15 g/L MSA and 50 ppm Cl−. Meanwhile, 4 ppm bis-(3-sodiumsulfopropyl disulfide) (SPS), 300 ppm polyethylene glycol (PEG) and 20 ppm Janus Green B (JGB) were used as the accelerator, suppressor and leveler, respectively. The MSA electrolyte solution which contained all of them was used to electrodeposit copper in via. The electrolyte was activated at 0.01 ASD for 5 min first, followed by filling which is performed at 4 mA/cm2 for 2 h. After copper filling, cross-section of samples was observed by FE-SEM (Field emission scanning electron microscope, sirion 200, FEI). Vacuum cold inlaying machine was used to decrease the air pressure to 10 Pa in cell (KC-B, Shenyang sysile machine factory). The chip sample is cut into property size and adhered to the copper sheet in right position. And the device of copper filling is shown in Fig. 1. The chip is linked to the galvanostatic power before immersing the MSA electrolyte solution to perform the copper filling.

3 Results and discussion

Taking the function of pretreatment into consideration, BES which is an acidic solution was chosen as the pre-wetting liquid in pretreatment. Wetting characteristics of it was measured both on copper oxide layer and copper layer, considered that copper oxide film exists on the copper seed layer under some conditions [22, 23]. The microscopic images of droplets formed on the two kinds of surface are shown in Fig. 2. It is evidently illustrated that wetting property of BES obviously decreases on the copper seed layer after copper is oxidized, demonstrated that contact angle from 59° to 84°. Therefore, copper oxidation on the surface of seed layer is a serious problem as the hydrophilic transfers to hydrophobic between the surface of seed layer and pre-wetting liquid, which means the force is necessary for making the completed contact. And the mechanism of that is explain in our former research [19].

Fig. 2
figure 2

Contact angle between BES: a copper seed layer; b copper oxide layer

Full size image

According to the results of wettability measurement, BES as a pre-wetting liquid is utilized in pretreatment, and the chip with TSVs is immersed in the BES for 5 min after vacuuming. Then, the chip with TSVs is put into plating bath for copper filling. Figure 3 shows partial filling of TSVs resulting from 1 h of deposition at current density 4 mA/cm2 in the electrolyte containing 4 ppm SPS and 300 ppm PEG. Effect of pretreatment is evident in the substantial difference of deposit thickness within TSVs, with or without pretreatment. With pretreatment, the copper deposited continuously on the copper seed layer and “superfilling” is forming during the partially deposited time as shown in Fig. 3a. In contrast of that, the copper is not deposited on the lower section of via under same deposited condition as shown in Fig. 3b. The comparatively results of copper deposit on the lower section of TSVs where electrolyte fails partway down suggests in these regions is dominated by the effect of pretreatment.

Fig. 3
figure 3

The cross-section images of copper filling at 4 mA/cm2 for 1 h: a BES served as a wetting liquid in pretreatment; b without pretreatment

Full size image

Meanwhile, when the copper seed layer is oxidized, the effect of pretreatment with BES is also investigated. In order to form copper oxidation on the surface of seed layer, the chip with vias has been put into chamber at 150 °C for 0.5 and 1 h, with considering that the copper seed layer is oxidized slightly and severely. In contrast with the original sample, the sample with copper oxidation on the surface of seed layer is examined by Energy Dispersive Spectrometer (EDS) as shown in Fig. 4. It shows the element of seed layer results of 30 μm in diameter and 100 μm in depth. Element distribution through EDS line scanning form via top to via bottom. Based on the EDS analysis of seed layers, it is found that oxygen distributes on the whole surface of seed layer. The content of oxygen increases with oxidation time increases in Fig. 4b, c, but it is not acceptable to measure the thickness of copper oxidation layer by EDS. The background element is demonstrated in results because magnitude of EDS depth (1 μm) is bigger than that of seed layer thickness (150 nm). It is not investigated fully in paper because our research focuses on the copper oxidation. The result shows that the copper on the surface of seed layer is oxidized and measurement of oxidation thickness is interpreted in next section.

Fig. 4
figure 4

EDS results of copper seed layer with different oxidized time which is demonstrated on the cross-section of TSV: a 0 h; b 0.5 h; c 1 h

Full size image

If copper oxidation exists on the surface of seed layer, the force, which is supplied by atmospheric pressure, is necessary to make pre-wetting liquid closely contact with seed layer as shown in Fig. 5. The pretreatment with pre-wetting liquid and vacuum has been proposed, which demonstrates the ability to remove the air bubbles in via. The different sample with different oxidation layer is filled under the same electrodeposited condition which is introduced in experiment section. Compared with the copper filling with normal copper seed layer (without oxidation), the copper electrodeposited layer is also continuous when copper seed layer is oxidized for 0.5 h as shown in Fig. 6a, b. Figure 6c shows that copper seed layer completely disappeared from the partial sidewall of via when copper seed layer is oxidized for 1 h in chamber. Considered that the copper seed layer is not uniform when the “scallops” structure exists on the sidewall of via, the copper seed layer is thinner at the top of “scallops” structure and it is corroded by acid [2, 20, 24, 25]. As shown in Fig. 6c, d, the edge of the coating gap is different from the typical shape of copper fracture caused by mechanical [26–28]. The copper is not deposited on the lower section of via because the cooper seed layer not exists on this section of via. It demonstrates that the copper seed layer is destroyed after oxidation with the pretreatment process. In contrast with shorter time of oxidation process in Fig. 6b, the copper of seed layer is about entirely oxidation on the lower section of via in Fig. 6c. It means that the thickness of copper seed layer at the bottom of via is thinner than that on the top of via, which is caused by fabrication of via and seed layer. Therefore, it is inferred that copper seed layer is destroyed by BES when it is used as the pre-wetting liquid during pretreatment when copper seed layer is oxidized completely on the sidewall of via.

Fig. 5
figure 5

Illustration of pretreatment mechanism with pre-wetting liquid and vacuum if the copper oxidation layer exists on the copper seed layer

Full size image
Fig. 6
figure 6

SEM images of copper filling while MSA electrolyte serves as a wetting liquid in pretreatment: a 0 h; b 0.5 h; c 1 h; d zoom image for (c)

Full size image

In order to further investigate the corrosion capability of BES, the corrosion experiment of copper seed layer was performed in BES. Copper seed layer on the surface of chip with TSV was partially corroded to form a step by 30 % nitric acid solution. Then the thickness of copper seed layer was measured by Probe profilometer. The original thickness of copper seed layer is 155 nm on the surface of chip. To quantitively evaluate the corrosion behavior of BES, the chip was corroded in solution for different time. All sample points were taken from the surface of chip. The thinning result of copper seed layer is shown in Fig. 7. It is calculated that the corrosion rate of copper in the BES is 1.36 × 10−2 nm/s at first soaking term (1 h). Combined that the whole pretreatment time is 5 min, 1-nm copper seed layer will be dissolved according to the corrosion rate. Therefore, pretreatment with BES is no harmful to corrode copper seed layer under normal condition.

Fig. 7
figure 7

The thickness of copper seed layer after soaking in BES for different time

Full size image

The thickness of copper oxidation layer is measured by electrochemical method and calculated by Faraday’s law [29],

$$ \updelta = \frac{it}{nF}V $$
(1)

where δ is thickness of copper oxide layer; i is current density (A/cm3) and its value is 50 mA; t is time; F is Faraday’s constant; V is molar volume of copper oxide; n is electron number which is decided by transferring electron of copper oxide [29]. Figure 8 shows that cathodic chronopotentiograms recorded for four different conditions of copper oxidation. Because copper oxide is dissolved quickly in BES solution, the accurate thickness of copper oxidation is hard to acquire in experiment when the oxidized time is 0.5 h. Hence, there is only the measured result of copper oxidation for 1 h in chamber. There is reduction of the electrode reaction at 300 s, as shown in Fig. 8 curve [1]. After soaking in BES for 5 s, the reversibility of the electrode disappeared. The copper oxide film is completely dissolved in 5 s as shown in Fig. 8. Table 1 shows the thickness of copper oxide film measured by X-ray Fluorescene thickness measurement and Probe Profilemeter. Each result is an average value of five sample points on the copper oxide film. Combined with the results of other measurements, a conclusion is obtained that the copper oxide film consists of cupric oxide and thickness of the film is 58.2 nm. Compared with curve [1] and curve [2], it is obvious that the copper oxide layer is completely dissolved after soaking in BES for 5 s, as shown in Fig. 6. Calculation demonstrates the corrosion rate of copper oxide is about 11.64 nm/s, which is up to three orders of magnitude greater than that of copper in BES (1.36 × 10−2 nm/s as stated above). It indicates that the copper seed layer is damaged much quicker after oxidizing in acidic electrolyte.

Fig. 8
figure 8

Results of polarization resistance method about corrosion of copper oxide layer after soaking in the BES for different time

Full size image
Table 1 Results of thickness of copper oxide films on the surface of chip (unit nm)
Full size table

4 Conclusion

In this paper, pretreatment including pre-wetting liquid and vacuum is utilized for assuring copper filling with viod-free. Due to the big contact angle between BES and copper oxide layer, BES was investigated with atmospheric pressure for its functionality, such as pre-wetting and cleaning, when the copper seed layer is oxidized to copper oxide on the surface. In addition, the corrosive rate of copper oxide is three orders magnitude greater than that of copper in BES. Hence, it is a disaster for copper filling that native copper grows on the surface of seed layer because copper oxide not only damages the function of seed layer but also accelerate the corrosion rate of copper seed layer in acidic electrolyte solution. The results demonstrate that BES is a desirable wetting agent for copper filling because it can sufficiently corrode the copper oxide on the surface of seed layer without damaging the copper seed layer during pretreatment.