I. INTRODUCTION

As semiconductor devices become more and more miniaturized, new semiconductor manufacturing processes are being used and new materials are being introduced. However, the use of polysilicon, which has been used as a material for semiconductor devices for a long time, is also increasingly used. The polysilicon gate has been replaced by a metal gate because it is difficult to satisfy in technology nodes over 45 nm, and the metal gate is fabricated with a last gate process. One of the key technologies in this process is to remove the dummy polysilicon by wet chemical ^(1-3).

In the case of Nand Flash Memory, there is a limit to high density on the flat panel, and the development is proceeding in 3D structure. This vertical NAND device with a polysilicon channel that overcomes scaling limits and process complexity, and achieves ultrahigh-density memory devices. One of the key technologies in this process is to maintain uniform thickness of polysilicon with mirror-like surface after wet chemical etching ^(4-9).

There are basically two types of etching solution of polysilicon; acid and base type. First, the mechanism of acid type is that nitric acid oxidizes the surface of silicon and hydrofluoric acid decomposes silicon oxide. We can easily control the etch rate by controlling the composition ratio of these, and the etch rate can modulate fast enough. The acid type is therefore used as a silicon etchant in a variety of applications and processes. However, these compositions are difficult to use in the presence of silicon oxide and silicon nitride ⁽¹⁰⁾. In this case, ammonia chemistries such as Tetramethyl ammonium hydroxide (TMAH), Tetraethyl ammonium hydroxide (TEAH), Potassium hydroxide (KOH) and diluted ammonia water can be used to etch ^(3,11). These hydroxide ions have been continuously studied to increase the selectivity of silicon to silicon oxide ^(12,13). And recent research has focused on the selectivity ratio between the crystal plane of monocrystalline silicon and the polysilicon ^(14,15).

These have recently been applied to deep trench etching in Micro electro mechanical system (MEMS) or Through silicon via (TSV) processes ^(16-19). Their mechanism consists of three steps. In the first step, hydroxide ions are diffused and adsorbed to the silicon surface. In the second step, the silicon is converted into silicon hydroxide and dissolved in water. Finally, hydrogen is produced in the reaction step of the hydroxides. Acid type has an advantage in etch rate at full back, but base type is advantageous when half-back or selectivity is required ^(11-15).

The trend in the semiconductor technology is to increase component density and reduce device minimum dimensions, such as channel length and gate oxide thickness. These changes make devices more susceptible to gate oxide related failures, such as shorts or leakage due to gate oxide holes or soft gate oxide breakdown. As a solution to this problem, semiconductor wet process is changing from dipping process to single tool process. This is because the single tool process is more advantageous for particle management than the dipping process. However, this single tool process differs greatly from the dipping process in the external influences to which the chemical is applied. In particular, there is a difference in oxygen concentration. The dipping process is a method of injecting chemicals into a vessel and dipping and unloading wafers, but a single tool process continuously sprays chemicals while the wafer is spinning or puddling. Therefore, the single tool process can increase the dissolved oxygen by widening the contact area where the chemical meets with the atmosphere. The influence of dissolved oxygen in the cleaning process has been studied in Hydrogen fluoride (HF) chemical ^(20,21).

In this study, we applied a half-back etching condition using hydroxides as an etchant. And we investigated the effect of dissolved oxygen on etching process in single tool process. In addition, we investigated the change of polysilicon surface and etch characteristics according to the etching condition and the wetting tendency of the chemical when etching the polysilicon.

II. EXPERIMENT

The polysilicon wafer to be used for the experiment was fabricated by Low pressure chemical vapor deposition (LPCVD) and deposited at 620 $^{\circ}$C in the condition of 0.2 torr with 100% SiH$_{4}$. It was grown on a 12-inch silicon wafer to a thickness of 200 nm and the growth rate was 100 nm/min. The etching test was carried out using this wafer.

The etching test was conducted by using both wafer level test and coupon level test. The wafer level test was performed using a single tool. Single tool equipment was evaluated using a 12-inch wafer using ZEUS's APOLLON model. First, 100:1 diluted HF (DHF) was pre-treated at 500 rpm for 60 sec subsequently rinsed with Di-ionized water (DIW) at 500 rpm for 30 sec, and spin-dried at 1500 rpm for 60 sec. After that, experiments were carried out. The etching test was carried out for 60 sec at 60 $^{\circ}$C by varying rpm condition. DIW rinse was herein performed at 500 rpm for 30 sec and then spin-dry was run at 500 rpm for 60 sec. Second, the coupon level test was performed by cutting polysilicon wafers into 1cm by 1cm specimens and dipping them in 100:1 DHF solution for 60 sec. Thereafter, rinsing was carries out with DIW for 30 sec and the resultant was blown with nitrogen gas. After the pretreatment was completed, 200 ml of the prepared test solution was filled in a 250 ml glass beaker, and the specimens were dipped in the test solution for 60 sec, were dipped in the DIW for 30 sec, and then were blown with nitrogen gas. Dipping conditions were either dipping only or adjusting the rpm using a magnetic stir bar.

The film thickness of the polysilicon before the chemical treatment and the film thickness after the chemical treatment were measured using an ellipsometer. The etch rate was calculated by two values. The equipment used in this experiment was the FE-5000S model of Otsuka Electronics. The light source used was a Xe lamp and its thickness was measured using a wavelength range of 300 to 800 nm. The Atomic force microscope (AFM) experiment for Root mean square (RMS) measurement was performed using a Nano-Focus’s n-Tracer, and the scan rate was 0.5 Hz and the scan size was 5 μm ${\times}$ 5 μm.

Dissolved oxygen was measured using an Origin Star DO meter instrument from Thermo Electron Corporation. First, the equipment was calibrated in DIW. The dissolved oxygen was measured after immersing the electrode in the test solution. One minute after the experiment, the dissolved oxygen values displayed on the display were recorded. Dissolved oxygen was measured according to the stirring rpm. The contact angle was measured using a Kruss DSA100 model and a 5 cm ${\times}$ 1 cm size polysilicon surface was measured. The test solution was sampled using a 50 cc syringe and set in the instrument. Then, test solution was dropped on the surface of the polysilicon to measure the contact angle with respect to the surface of the solution and polysilicon. Raman spectroscopy was measured using NTEGRA Spectra NT-MDT. The beam expander wavelength was 633 nm and the exposure time was 10 sec.

III. RESULT & DISCUSSION

Experiments were conducted to investigate the behavior of polysilicon etching depending on the kind and content of the compound having a hydroxide ion. TMAH, TEAH and Choline hydroxide having hydroxide ions were compared each other. This test is to compare the content of hydroxide ion and kinds of cation. The etching test was carried out after removing the native oxide film for 30 sec by 100:1 DHF solution on 12 inch wafer rotating at 500 rpm in a single tool. The etch rate and surface roughness were measured according to the concentration of each alkali compound. The thickness of the polysilicon film was measured with an ellipsometer, and the surface roughness was measured using AFM. The results are shown in Table 1.

Compared with TEAH and choline hydroxide, TMAH showed higher etch rate. It can be seen that the etch rate does not increase continuously in proportion to the concentration of the hydroxide but the etch rate converges to some extent at the concentration of the hydroxide ion. In the case of TEAH, the etch rate is slower than TMAH. This result is expected to be due to the poor penetration of polysilicon because TEAH has a longer carbon than TMAH. It has a similar RMS value compared to TMAH and a slower etch rate, making it difficult to apply as a basic chemical in polysilicon etching. In the case of choline hydroxide, the etch rate is slower than TMAH but the RMS value is better. It is not clear whether the RMS value is improved due to the late etch rate, or because the hydroxide ion in the chemical structure has a high adsorption force on the polysilicon surface, thus preventing hydrogen from the etching process from adsorbing on the polysilicon surface. However, choline hydroxide also does not have sufficient etch rates, making them difficult to apply as basic chemical in polysilicon etching process. In the case of TMAH, etch rate increased with increasing content, but saturation occurred with more than 10wt.%. Increasing the content of TMAH, TEAH, and Choline hydroxide results in saturation of the etch rate due to the decrease in water. The -OH produced in water affects the etching rate, and it seems that the etching rate does not increase more than the water content decreases to some extent. And the RMS also shows the same tendency, which is not clear whether it is due to the concentration of hydroxide ion or proportional to the etch rate. If the direction of increasing the etch rate is the direction that does not improve the RMS, we need to discuss how to improve it. This is because the RMS must be good in the half-back etching process.

Two methods are considered feasible; the first is to treat the surface. This is because, as described above, hydrogen is generated in the third step of the poly-etching mechanism by the hydroxide ion ^(11-15). If the hydrogen gas is not discharged directly out of the system, it is adsorbed on the surface of the polysilicon by bubbles. This bubble can serve as a mask in the etching process and can make the surface roughness of the polysilicon worse. Surfactants can therefore be applied to improve the RMS while maintaining the etch rate. In addition, the surface tension of the etching solution may be expected to be improved by the surfactant.

In order to understand how surface tension affects contact angle, RMS and the etch rate. The change of these characters was measured as a function of surfactant content as shown in Fig. 1. Herein, the content of TMAH was fixed at 7 wt% and the addition of MEGAFACE F-281 surfactant was varied. The test was carried out using the single tool equipment as before.

These results implied that the contact angle and the RMS were decreased by surfactant effect and the hydrogen bubble is prevented from adsorbing to the surface. The question is how to reduce RMS value concomitance with increase etch rate.

Eqs. (1, 2) are the chemically etched reaction of the surface-oxidized silicon and silicon with hydroxide, respectively. Eq. (2) is the etch reaction of silicon with hydroxides. The slowest step in these reactions is the reaction step where the silicon surface becomes Si-OH by -OH. This step determines the reaction rate. The dangling bonds on the silicon surface are first replaced with hydrogen to become H-terminated silicon. H-terminated silicon has higher density than silicon dioxide, resulting in low surface stability and strong electronegativity between atoms. As a result, the silicon atoms are subjected to a strong attack of nucleophiles. Therefore, it is known that the reaction rate of Eq. (2) is 100 to 1,000 times faster than that of Eq. (1) ⁽²²⁾.

Therefore, we have developed a method to increase etch rate of polysilicon and to examine the process. It was found that the etch rate of the single tool was lower than that of the dipping ^(20,21). It was attributed to dissolved oxygen in the chemical. It is difficult to etch by TMAH if polysilicon becomes silicon dioxide by dissolved oxygen. Therefore, we have examined whether dissolved oxygen affects the etch rate of polysilicon in a process using single tool equipment. In order to carry out the simulation evaluation, a composition was prepared by adding 7 wt% of TMAH and 100 ppm of the above surfactant to a 250 ml beaker and filling 200 ml thereof. We measured the dissolved oxygen while changing the rpm of the stir bar. We could continue to test how the etch rate and RMS would change if we could reduce the dissolved oxygen. So, we applied Di-ethyl hydroxyl amine (DEHA), which is considered as an oxygen scavenger. It is known that DEHA will acts as an oxygen scavenger at operated temperature between room temperature and 80 $^{\circ}$C by followed Eq. (3) ⁽²³⁾.

First, 0.1, 0.5 and 1.0 wt% of DEHA solution was prepared in the basic composition of TMAH 7 wt% and surfactant 100 ppm. Then, 200 ml of sample was placed in a 250 ml beaker and the dissolved oxygen was measured while controlling the rpm with a stir bar. At the same time, the etch rate was measured by dipping a polysilicon 1 cm ${\times}$ 1 cm specimen for 1 min.

Fig. 3 shows that the dissolved oxygen of about 5 ppm is included in the composition under the condition of not stirring, but the dissolved oxygen increases to 7.5 ppm in the stir condition. As the rpm increases, the dissolved oxygen does not increase further. It seems to be saturated. The etch rate significantly decreases at 200 rpm, as does dissolved oxygen, and then decreases slowly thereafter. However, the etch rate decreases at 200 rpm and gradually decreases with increasing rpm. At 300 rpm, the etch rate varies depending on the content of DEHA, but the difference in etch rate according to the DEHA content is insignificant at 400 rpm. This shows that although DEHA is effective in inhibiting dissolved oxygen up to 300 rpm, it is difficult to effectively remove dissolved oxygen when it is above 400 rpm. As the rpm increases, the amount of dissolved oxygen in the beaker is expected to increase. This is because the amount of oxygen present in the atmosphere widens the area in contact with the chemical. In this composition, the etch rate is related to the dissolution rate of dissolved oxygen; as the dissolved oxygen content increases, the etch rate decreases. This is expected to affect the single tool process. Although the dissolved oxygen is controlled, it does not seem to have a great effect at 400 rpm or higher. Therefore, it seems preferable to operate at 300 rpm or less.

However, Fig. 3 shows that there is a tendency for the influence of dissolved oxygen and etch rate to be inconsistent. The surface roughness was characterized using AFM to verify DEHA effect on the etch mechanism in addition to the removal of dissolved oxygen. Fig. 4 shows that the decrease in dissolved oxygen with increasing DEHA not only affects the etch rate of polysilicon but also minimizes the oxidation of the surface of the polysilicon while improving the surface roughness.

The DEHA effect on the etch surface was further investigated using Raman spectroscopy. Fig. 5(a) exhibits Raman spectra for etched polysilicon samples with different etching conditions. It shows the difference in the surface of polysilicon treated with 7 wt% solution of TMAH and the mixture of 7 wt% of TMAH and DEHA with 1 wt% solution of polysilicon without chemical treatment. Fig. 5(b) shows the interval spectra between 700 cm$^{-1}$ and 1200 cm$^{-1}$ to exclude the intensity difference caused by the increase of the roughness due to etching of polysilicon.

Line (a) is a Raman spectrograph of the surface of polysilicon cleaned with DIW after etching polysilicon with 7 wt% TMAH, and line (b) is a Raman spectrograph of the surface of polysilicon etched polysilicon with 7 wt% TMAH without DIW cleaning. Line (c) shows a Raman spectrograph of DIW cleaning after etching the surface of polysilicon with a mixture of 7 wt% TMAH and 1 wt% DEHA. Line (d) is the polysilicon surface without any chemical treatment. Polysilicon (line (d)) without treatment and polysilicon (line (a)) cleaned after TMAH treatment alone show 867 cm$^{-1}$ peak. So, this is seen as the peak of etching reaction by TMAH. However, line (c) shows a peak at 1043 cm$^{-1}$ when not washed. This is also seen in line (c), where a peak at 1043 cm$^{-1}$ can be deduced from C-N and Si-N. In line (c), there is a peak of 956 cm$^{-1}$ referred as to the peak of Si-N and Si-O-N ^(24,25).

It can be seen that DEHA plays a role of reducing the oxygen concentration by capturing oxygen and also contributing to the roughness of the polysilicon surface by participating in the reaction. This is an important role in the mechanism of polysilicon etching with alkaline chemicals, especially, for single-tool processors that are very sensitive to oxygen concentrations.

IV. CONCLUSION

As the etching process for polysilicon is changed from a dipping process to a single-tool process, etching with alkaline material is affected by dissolved oxygen. Dissolved oxygen causes poor surface roughening and low etch rate. In order to improve the surface roughness, the surfactant usually utilized but the etch rate is further lowered. The DEHA was herein applied not only to reduce the dissolved oxygen but also to increase the etch rate. Our results indicated that DEHA facilitates the etch reaction on the polysilicon surface and reduces the surface roughness.