Effect of C₂H₂F₄/CF₄O with low global warming potentials on SiN_x etching as a CHF₃ replacement

Abstract

As the size of the device continues to decrease due to the increasing demand for faster processing speed and lower power consumption in semiconductor integrated circuit device technology, the double patterning process is widely used. The SiN_x used in this double patterning process requires high etch rate and high etch selectivity over SiO_x, while at the same time achieving an anisotropic etch profile. In the past, gases such as CHF₃/CF₄ were used for SiN_x etching of double patterning. However, the low etch selectivity of these gases and their high global warming potential (GWP) have led to the need for alternative gases. To address this issue, in this study, the effect of alternative gases instead of CHF₃ on SiN_x etching characteristics has been investigated. When C₂H₂F₄ was used instead of CHF₃, both etch rate and etch selectivity were improved, but issues such as trenching and increased critical dimension (CD) were observed. When CF₄O was added to C₂H₂F₄, both etch rate and etch selectivity were further improved while eliminating trenching issue. The analysis showed that C₂H₂F₄ compared to CHF₃ promoted stronger polymer formation, thereby improving mask passivation while trenching defects occurred due to polymer deposition. For the C₂H₂F₄ + CF₄O mixture, increased fluorine dissociation resulted in higher SiN_x etch rates and consequently better etch selectivity, and trenching was eliminated by increasing gas dissociation and decreasing polymer production. Million Metric Tones of Carbon Equivalent (MMTCE) measurements showed that C₂H₂F₄ decreased by approximately ~83.9% and C₂H₂F₄ + CF₄O by approximately ~75.2% of greenhouse gas emissions compared to CHF₃. Therefore, the results obtained with alternate gases are next-generation eco-friendly etch processes that can be applied to semiconductor devices such as Fin Field Effect Transistor (FinFET), 3D Not-AND (NAND), and other advanced semiconductor and display manufacturing technologies.

Introduction

Today, SiN_x and SiO_xoffer high performance at low cost with excellent insulating properties and high breakdown strength^1,2,3,4,5. These advantages have led to their widespread use in applications such as Fin Field-Effect Transistors (FinFETs) and Not-AND (NAND) that require high electrical performances^{6,7,8,9,10,11}. In recent years, the size of devices has been steadily decreasing due to the continuous increase in demand for faster processing speeds and lower power consumption, and, to reduce device sizes below the photolithography wavelength limit, double patterning techniques are widely used^12,13. The double patterning process requires a highly selective etching process by using an insulated upper hard mask instead of amorphous carbon layer (ACL) or bottom anti-reflective coating (BARC) layer^{2,14,15,16,17}. This allows for the removal of unnecessary bottom layers through high etch selectivity, reducing process time and double patterning defects, leading to cost savings. In particular, increasingly complex and precise semiconductor etch processes require deeper SiN_x etching than ever before, requiring high etch selectivity between SiN_x and SiO_x^{18,19,20,21,22,23,24,25}. This is because employing a highly selective etching process between SiN_x and SiO_x can effectively eliminate the multiple etching steps that are typically required in conventional double patterning techniques. Therefore, the objective of this study is to enhance process efficiency through the implementation of highly selective etching between SiN_x and SiO_x, which removes the necessity for multiple etching sequences commonly employed in traditional double patterning approaches.At present, SiN_x is etched using gases with high global warming potential (GWP) gases such as NF₃/O₂, CHF₃/O₂/Ar, CHF₃/CH₃F/CH₂F₂/Ar, N₂/O₂/CF₄, SF₆/H₂/He/Ar as etchants^16,26,27,28. However, CHF₃, CF₄, SF₆, and NF₃have high GWPs. With the development of next-generation etch processes, these high GWP gases must be replaced by lower GWPs^29,30,31,32. To address this issue, a relatively low GWP gas, C₂H₂F₄ (HFC-134a), is being considered as a potential alternative to CHF₃. However, it is still high and needs further improvement.In this study, the applicability of C₂H₂F₄ to SiN_x etch process as a replacement of CHF₃ using inductively coupled plasma (ICP) during the double patterning process that requires selective etch compared to SiO_x and highly anisotropic etch profiles has been investigated, and, in addition, the effect of CF₄O addition to C₂H₂F₄ on the SiN_x etch characteristics has been studied. Because CF₄O helps suppress excessive polymer formation and improve profile control, it was added to C₂H₂F₄. Its low GWP, combined with its ability to increase F and O radicals in the plasma, makes it an effective additive for enhancing SiN_x etching performance while reducing environmental impact. First, it was shown that replacing CHF₃ with C₂H₂F₄ improves etch selectivity over SiO_x; however, it also resulted in an undesirable increase in critical dimension (CD) and the formation of etch profile defects such as trenching, which are considered negative effects in terms of pattern fidelity and process reliability. However, the addition of CF₄O solved these etch issues, further improving etch rate and selectivity. In addition, Fourier transform infrared spectroscopy (FT-IR) analysis showed that SiN_x etch processes with C₂H₂F₄ containing CF₄O improved Million Metric Tons of Carbon Equivalent (MMTCE) values when etching SiN_x of the same thickness compared to conventional CHF₃ processes.

Experimental

The 300 mm inductively coupled plasma (ICP) system used in the study is shown in Fig 1(a). The ICP source consists of an antenna consisting of two copper coils, one inside and one on the outside. A ~35mm thick alumina window was installed above the electrode to pass the electromagnetic field formed from the antenna into the chamber, and in addition to a gas ring located at the top edge of the chamber wall for injecting gas from the side, a gas hole was made in the center of the alumina window for uniform gas distribution in both the center and the edge. The substrate was placed on the powered lower electrode of the ICP system, which was equipped with active cooling using a chiller system. The distance between the substrate surface (placed on the powered lower electrode) and the ICP source region was approximately ~15 cm. The ICP power to the antenna is delivered by a 13.56 MHz radio frequency (RF) generator (Seren-R3001), and the bias power to the substrate is supplied by a 2 MHz RF generator (Seren-R2001). To regulate the process pressure, the chamber is equipped with an automatic process pressure controlling pendulum valve (VAT model PM.7) between the turbopump and the dry pump. At the downstream of the dry pump, an FT-IR (MIDAC, I2000) was installed to measure the recombination gas species. The experiment was performed with a sample of SiN_x (200nm) patterned with SiO_x (110nm) on a silicon wafer. The critical dimension (CD) of the SiO_x hardmask was ~73 nm and the space between patterns was ~71 nm. A SEM cross-sectional image of SiO_x patterned on SiN_x is shown in Fig 1(b). The SiO_x patterned SiN_x etch process was performed under the conditions of an ICP source power of 1500 W with 13.56 MHz RF power, a DC bias of −100 V with the 2 MHz RF power, a process pressure of 4 mTorr, and a substrate temperature of 18 °C. The total gas flow rate was maintained at 340 sccm, consisting of CH₃F (40 sccm), O₂ (90 sccm), and He (100 sccm) supplied in common, together with either CHF₃ (110 sccm), C₂H₂F₄ (110 sccm), or a mixture of C₂H₂F₄ (80 sccm) + CF₄O (30 sccm) as the main etchant gas. The etch rate and selectivity, and the etch profile of SiO_x patterned SiN_x were examined using the field emission scanning electron microscope (FE-SEM; Hitachi, S-4700), and SEM was also used to measure CD. The plasmas produced by the dissociation of hydrofluorocarbon (HFC) and perfluorocarbon (PFC) gases were observed by optical emission spectroscopy (OES; ANDOR, Technology SR-ASZ-0103). In addition, radicals and cations were measured using quadrupole mass spectrometry (QMS; Hiden Analytical, PSM 500). The surface composition and binding states were measured by X-ray photoelectron spectroscopy (XPS; VG Microtech Inc. ESCA2000). Fourier transform infrared spectroscopy (FT-IR; MIDAC, I2000) was used to measure the molecular species emitted outside of the process chamber and to calculate the total global warming potential of the process gases, Million Metric Tons of Carbon Equivalent (MMTCE) values, used in the experiment.

Fig. 1

(a) Schematic diagram of the 300 mm ICP system used in the experiment. (b) a SEM image of SiO_x masked SiN_x/Si wafer structure used in the experiment.

Results and discussion

SiN_x/SiO_x etching

Fig 2 (a)-(c) show the etch rates and etch selectivity of SiN_x over SiO_x for each etch gas such as (a) 110 sccm CHF₃, (b) 110 sccm C₂H₂F₄, and (c) 80 sccm C₂H₂F₄+ 30 sccm CF₄O measured as a function of O₂ flow rate. The process conditions were source power 1500 W, DC bias −100 V, process pressure 4 mTorr, and substrate temperature 18° C. CH₃F (40 sccm) and He (100 sccm) were commonly supplied to the chamber. In the case of CHF₃, as shown in Fig 2 (a), etch rates of both SiN_x and SiO_x decreased with increasing O₂ flow rate possibly due to the decreasing etching species in the plasma but the SiN_x etch rate decreased faster possibly due to the oxidation of SiN_x surface, therefore, the etch selectivity of SiN_x over SiO_x decreased with increasing O₂ flow rate and the highest etch selectivity of ~ 3 was observed at 30 sccm O₂ flow rate. In the case of C₂H₂F4, as shown in Fig 2(b), the etch rates of both SiN_x and SiO_x increased with increasing O₂ flow rate. Due to the faster increase of SiO_x etch rate with O₂ flow rate, the etch selectivity of SiN_x over SiO_x was decreased. The further increase of etch rates of SiN_x and SiO_x with increasing O₂ flow rate appears to be related to the removal of a fluorocarbon polymer layer formed by carbon-rich C₂H₂F₄ by increased O in the plasma. When 30 sccm CF₄O was mixed with 80 sccm C₂H₂F₄ as shown in Fig 2 (c), the etch rates of both SiN_x and SiO_x were higher than 110 sccm C₂H₂F₄ at 30 sccm of O₂ possibly due to lower fluorocarbon polymer on the surfaces of materials. And the further increase of O₂ flow rate did not change the SiN_x etch rate significantly while decreasing SiO_x etch rate slightly, therefore, the etch selectivity of SiN_x over SiO_x was slightly increased with increasing O₂ flow rate. (Additional optimization results are presented in the Supplementary Information Figure S1. Although the highest etch rate was observed under one condition, this setting caused reduced SiN_x/SiO_x selectivity and profile degradation. Therefore, the condition that provided a better balance among etch rate, anisotropy, and selectivity was chosen as the optimized condition.) As a result, the SiN_x etch rate and the etch selectivity of SiN_x over SiO_x were the highest as ~ 160 nm/min and ~ 4, respectively, at 30 sccm CF₄O + 80 sccm C₂H₂F₄. Therefore, it was found that, for the etching of SiN_x selective to SiO_x, CHF₃-based gas can be replaceable by C₂H₂F₄ and, especially, by C₂H₂F₄+CF₄O with higher SiN_x etch rate with higher etch selectivity over SiO_x. Also it should be noted that the etch rate and selectivity improvements observed in this study are not solely due to the use of low GWP gases (C₂H₂F₄ and CF₄O). The addition of O₂, He, and CH₃F also plays an essential role: O₂ reduces excessive polymer formation, He stabilizes plasma and enhances etch uniformity, and CH₃F contributes to controlled sidewall passivation. These gases, together with the low GWP etchants, collectively determine etch characteristics. To confirm repeatability, the etching experiments were repeated on three separate wafer batches, and the variation in SiN_x/SiO_x etch selectivity was within ±3%, indicating high process reliability.Fig 3 (a)~(c) show SEM cross-sectional images of SiO_x masked SiN_x etch profile after etching ~ 50 nm deep SiN_x, which is generally required for SiN_x etch depth in conventional double patterning process. Etch gases were (a) CHF₃ (110 sccm), (b) C₂H₂F₄ (110 sccm), and (c) C₂H₂F₄(80 sccm)+CF₄O (30 sccm) and the other process conditions were the same as those in Fig 2. As shown in Fig 3(a), for SiN_x etching with CHF₃-based gas, SiO_x mask was consumed the most among three gases in (a)~(c) also, slight SiN_x sidewall etching was observed. When C₂H₂F₄ or C₂H₂F₄+CF₄O was used instead of CHF₃, as shown in Fig 3(b) and (c), respectively, highly anisotropic SiN_x etch profile with the smaller SiO_x mask etching than CHF₃ in (a) could be observed. However, in general, for the etching of ~50 nm deep SiO_x masked SiN_x etching, three etch gases were applicable for current double patterning process.

Fig. 2

Etch rate and SiN_x/SiO_x ratio as a function of O₂ flow rate; (a) CHF₃ (110 sccm), (b) C₂H₂F₄ (110 sccm) and (c) C₂H₂F₄ (80 sccm)+CF₄O (30 sccm). (Additionally, 40 sccm CH₃F and 100 sccm He were added commonly).

Fig. 3

SEM images of 50nm depth partial SiN_x etch with (a) CHF₃ (110 sccm), (b) C₂H₂F₄ (110 sccm), and (c) C₂H₂F₄(80 sccm)+CF₄O (30 sccm). SEM images for 200nm depth SiN_x full etching with (d) CHF_3, (e) C₂H₂F₄, and (f) C₂H₂F₄+CF₄O. The other process conditions are the same as those in Fig 2.

Next generation double patterning process requires deeper SiN_x etching, therefore, the etch time was increased to etch ~200 nm thick SiN_x, and the results are shown in Fig 3 (d) CHF₃, (e) C₂H₂F₄, and (f) C₂H₂F₄+CF₄O^{13,16,28,33,34}. For etching with CHF₃-based gas, as shown in Fig 3(d), it was difficult to etch ~ 200 nm deep SiN_x while shrinking CD size significantly and removing SiO_x mask almost completely, due to the surface oxidation of SiN_x at 90 sccm of O₂ flow rate. (When the O₂ flow rate was decreased to 30 sccm, as shown in Supplementary Information Figure S2, full 200 nm deep SiN_x could be etched with CHF₃-based gas even though the CD size was decreased and SiO_x mask was almost removed similar that at 30 sccm O₂ flow rate.) In the case of C₂H₂F₄ and C₂H₂F₄+CF₄O, as shown in Fig 3 (e) and (f), respectively, ~ 200 nm deep SiN_x could be fully etched and almost vertical etch profiles were observed. However, in the case of C₂H₂F₄-based gas, the CD size was slightly increased and trenching was observed at the etched SiN_xpattern bottom edge possibly due to a polymer layer formed at the pattern bottom. This trenching behavior appears to be in good agreement with previous reports, which suggest that similar profile defects observed in fluorocarbon plasmas may result from localized polymer deposition^17,35. However, for C₂H₂F₄+CF₄O-based gas, vertical SiN_x etch profiles without CD variation and without trenching defects could be observed possibly due to the effective polymer layer removal on the pattern bottom area by addition of CF₄O. As shown in Fig 3 (d-f), replacing CHF₃ with C₂H₂F₄ improved the SiN_x/SiO_x selectivity. Moreover, the addition of CF₄O further enhanced the selectivity, which can be attributed to the increased availability of fluorine radicals required for SiN_x etching and the simultaneous suppression of excessive polymer deposition by oxygen radicals. These improvements demonstrate that CF₄O plays a critical role in achieving both anisotropic etching and higher selectivity. Additionally, the sidewall angle could not be measured for the CHF₃-based process due to the incomplete etching of the SiN_x layer. For the C₂H₂F₄-based process, the measured sidewall angle was approximately ~84°, while for the C₂H₂F₄+CF₄O-based process, it was around ~88°, confirming highly anisotropic and vertical etch profiles. The etch uniformity across the wafer was maintained within ±5%.

Plasma analysis

Using the process conditions in Fig 3, the dissociated species in the plasma were observed using QMS and OES. Fig 4 (a) and (b) show positive ion species in the plasma and their relative intensities, respectively, measured by QMS. (Total measured ion intensities were measured and the results are shown in Supplementary Information Figure S3 and the total ion intensities were the highest for CHF₃-based gas and the lowest for C₂H₂F₄-based gas possibly indicating the lowest plasma density for the C₂H₂F₄-based gas.) Fig 4(a) shows an overview of the positive ion species detected in the plasma for each gas chemistry, as measured by QMS. As shown in Fig 4 (b), polymerizing ion species such as CH₂F⁺, CHF₂⁺, and C₂H₂F₃⁺ were slightly more abundant in the CHF₃-based plasma compared to C₂H₂F₄ and C₂H₂F₄+CF₄O plasmas, which showed similar levels. In addition, the CHF₃ plasma exhibited the highest O₂⁺ ion intensity, whereas the lowest O₂⁺ intensity was observed with the C₂H₂F₄-based plasma, indicating a stronger oxidation potential in the CHF₃ condition. The largest amount of O₂⁺ in the plasma for CHF₃-based gas is believed to be related to the oxidation of SiN_x while the lowest O₂⁺ amount in the plasma for C₂H₂F₄-based gas is related to the trenching of the patterned SiN_x edge due to the polymer formed on the pattern surface. Etch species such as HF⁺ and CF₃⁺ were the highest for C₂H₂F₄+CF₄O and the lowest for CHF₃⁺, and which is believed to be partially related to the highest SiN_x etch rate for C₂H₂F₄+CF₄O and the lowest for the CHF₃.Fig 4 (c) shows the OES wide scan data for CHF₃, C₂H₂F₄, and C₂H₂F₄+CF₄O-based gas in Fig 3 and (d)~(f) show OES narrow scan data related to (d) F (703.7 nm), (e) O (777.4 nm), and (f) H (656.5 nm)^{30,31,36,37,38}. The peak intensities were normalized by He (501.6 nm) peak intensity. The intensity variation of O and H observed by OES were similar to the variation of those positive ion mass intensities (for O, CHF₃>>C₂H₂F₄+CF₄O>C₂H₂F₄ and for H, CHF₃>C₂H₂F₄>C₂H₂F₄+CF₄O) The highest O₂⁺ ion intensity was observed in the CHF₃-based plasma, which is associated with enhanced oxidation and suppression of polymer accumulation. In contrast, the lowest O₂⁺ level in C₂H₂F₄-based plasma results in reduced oxidation, promoting polymer formation and the occurrence of trenching. In addition, using OES, the variation of F peak intensity in the sequence of C₂H₂F₄+CF₄O>C₂H₂F₄>CHF₃ could be observed, and which is believed to be related to the SiN_x etch rate together with variation of CF₃⁺ and HF⁺ observed by QMS.

Fig. 4

(a) Positive ions in the plasmas by QMS formed using CHF₃, C₂H₂F₄, and C₂H₂F₄+CF₄O-based gas in Fig 3. (b) Intensities of each positive ion. (c) OES wide scan data for CHF₃, C₂H₂F₄, and C₂H₂F₄+CF₄O-based gas in Fig 3. OES narrow scan data related to (d) F, (e) O, and (f) H normalized by He.

Surface analysis & etch mechanism

The atomic composition of the etched SiN_x and SiO_x surfaces and the binding states of Si under the process conditions in Fig 3 were measured by XPS, which are shown in Fig 5. All plasma and surface characterizations in this study were conducted on blanket wafers. As such, sidewall-specific effects in patterned features are not directly assessed. To minimize surface contamination, the etched samples were vacuum-packed to reduce exposure to air and then transferred to the XPS chamber. The measurements were carried out under high vacuum conditions (~10^-9Torr). For the XPS analysis, the SiN_x and SiO_x were etched for 2 min. Fig 5 (a) and (b) show the atomic composition of the etched SiN_x and SiO_x surfaces, respectively. As shown in Fig 5(a), the SiN_x surface has the most C percentage for C₂H₂F₄-based gas, followed by C₂H₂F₄+CF₄O, and then CHF₃, while O percentage has the most for CHF₃, followed by C₂H₂F₄+CF₄O, and C₂H₂F₄. Therefore, it is believed that the SiN_x surface etched by CHF₃-based gas was oxidized due to the highest O percentage while the SiN_x surface etched by C₂H₂F₄ was polymerized due to the highest C percentage. In addition, the higher percentages of Si and N in the order of CHF₃, C₂H₂F₄, and C₂H₂F₄+CF₄O appear to be related to the etch rate of SiN_x. In contrast, the etched SiO_x surfaces, as shown in Fig 2(a)–(c), exhibit similar etch rates regardless of gas chemistry, suggesting that the underlying etching mechanism remains consistent across the different plasma conditions. The surface characteristics were further investigated by observing XPS narrow scan data of Si 2p for the SiN_x and SiO_x etched with CHF₃, C₂H₂F₄, and C₂H₂F₄+CF₄O. The results are shown in Fig 5(c) CHF₃, (d)C₂H₂F₄, and (e)C₂H₂F₄+CF₄O for etched SiN_x surface and (f) CHF₃, (g)C₂H₂F₄, and (h)C₂H₂F₄+CF₄O for etched SiO_x surface. As shown in Fig 5 (c~e), the SiN_x etched with CHF₃ showed the highest Si-O bonding (~103.6 eV)³⁹ while the SiN_x etched with C₂H₂F₄ and C₂H₂F₄+CF₄O showed the highest Si-N bonding (~101.8 eV)⁴⁰ indicating the oxidation of SiN_x surface by CHF₃, which results in low SiN_x etch rate. (Supplementary Information Figure S4 shows XPS C1s narrow scan data for SiN_x and SiO_x etched with CHF₃, C₂H₂F₄, and C₂H₂F₄+CF₄O. Both SiN_x and SiO_x surfaces etched with C₂H₂F₄ showed the highest C–F (~289.5 eV), C–CF (~287.5 eV), and C–CF₂ (~291.8 eV) bondings related to a fluorocarbon polymer layer, and C₂H₂F₄+CF₄O exhibited slightly lower intensities, while CHF₃ showed the lowest among the three gases. Therefore, SiN_x etched with C₂H₂F₄ exhibited the thickest polymer layer on the etched SiN_x surface, and which might be related to the trenching of etched SiN_x pattern edge.) In the case of SiO_x surfaces, as shown Fig 5 (f~h), Si-O bonding (~103.6 eV) was the highest binding peaks for all the three gases, therefore, no significant differences in SiO_x etch rate could be observed. (Supplementary Information Figure S5 shows the atomic percentages of SiN_x and SiO_x surfaces measured by XPS after etching using CHF₃ with 30 sccm of O₂ flow rate, and XPS Si 2p and C1s narrow scan data on SiN_x and SiO_x surfaces etched using CHF₃ with 30 sccm of O₂ flow rate. The use of 30 sccm O₂ flow rate for the etching of SiN_x and SiO_x with CHF₃ decreased the O percentage and Si-O bonding peak on the etched SiN_x surface similar to those for the SiN_x etched using C₂H₂F₄ and C₂H₂F₄+CF₄O with 90 sccm O₂ flow rate while no significant change was observed for the etched SiO_x surface similar to the surface etched with 90 sccm O₂ flow rate.)Fig 6(a)~(c) shows the potential etch mechanism of SiO_x masked SiN_x pattern etching under the process conditions in Fig 3. Fig 6(a) shows the etch behavior of SiN_x when using CHF₃ gas. During the etching process with CHF₃, surface oxidation occurs on SiN_x surface, resulting in significant decrease of SiN_x etching. Fig 6(b) shows the SiN_x etching process using C₂H₂F₄ gas. The low surface oxidation of SiN_x maintains etching, enabling full etching with a thickness of ~ 200 nm. However, the formation of a polymer layer in the SiN_x pattern due to C₂H₂F₄ leads to problems such as tapered profile and trenching. Fig 6(c) shows the etching mechanism under the C₂H₂F₄+CF₄O. The addition of CF₄O reduced polymer formation, resulting in higher etching rates and better selection ratios compared to C₂H₂F₄ alone, eliminating trenching within the SiN_x pattern. Consequently, the trenching and tapered profiles observed under C₂H₂F₄-based plasma are primarily caused by excessive fluorocarbon polymer deposition, which disrupts uniform ion flux toward the pattern bottom. The incorporation of CF₄O promotes the oxidation of carbon-based species, thereby reducing polymer thickness and improving etch anisotropy.

Fig. 5

Atomic percentages of (a) SiN_x and (b) SiO_x surfaces measured by XPS after etching using the process conditions in Fig 3. XPS Si 2p narrow scan data on etched SiN_x for (c) CHF₃, (d) C₂H₂F₄, and (e) C₂H₂F₄+CF₄O and XPS Si 2p narrow scan data on etched SiO_x for (f) CHF₃, (g) C₂H₂F₄, and (h) C₂H₂F₄+CF₄O.

Fig. 6

Schematic drawings of etch mechanism for SiO_x masked SiN_x etching with CHF₃, C₂H₂F₄, C₂H₂F₄+CF₄O-based gases. (a) for CHF₃-based gas, due to the SiN_x oxidation, the etching of SiN_x is very slow. (b) for C₂H₂F₄-based gas, due to the polymer formation, trenching is formed at the SiN_x pattern bottom edge. (c) for C₂H₂F₄+CF₄O, the polymer layer in the pattern is effectively removed and highly anisotropic SiN_x etching is obtained.

Measurement of million metric tons of carbon equivalents

Table 1 provides detailed information including the names of the gases used in the experiment, their molecular weights, GWP values and the Chemical Abstracts Service (CAS) values. CHF₃ exhibits a high global warming potential (GWP100_yr) value of 14,600, C₂H₂F₄ shows a GWP100_yr value of 1,430, and CF₄O represents a very low GWP100_yrvalue close to ~1⁴¹. However, since the gases used in etching have the potential to recombine and form by-products after decomposition during plasma discharge, it is essential to analyze the gas concentration in the exhaust line and calculate the Million Metric Tons of Carbon Equivalent (MMTCE), which corresponds to the total global warming index. Below is the equation for the MMTCE. The MMTCE was calculated using GWP100_yr, which refers to the integrated global warming potential over a 100-year time horizon, while Mi represents the total mass emission of HFCs and PFCs measured by FT-IR during the process⁴²

Table 1 Information on CHF₃, C₂H₂F₄, and CF₄O used in the experiment (e.g., gas name, molecular weight, GWP 100 years, and CAS number).

$$\text{MMTCE}=\sum_{{\varvec{i}}}\frac{12}{44}\times \frac{{{\varvec{M}}}_{{\varvec{i}}} \times {{\varvec{G}}{\varvec{W}}{\varvec{P}}}_{100{\varvec{y}}{\varvec{r}}}}{{10}^{9}}$$

Fig 7(a) shows the concentration (in ppm) of exhaust gases measured in the exhaust line using FT-IR during plasma generation using CHF₃, C₂H₂F₄, and C₂H₂F₄+CF₄O-based gas under the process conditions in Fig 3, and which shows the various recombinants of decomposed etched gases. In the case of CHF₃, recombinants such as HF (GWP = 0), COF₂ (GWP = ~1), and CHF₃ itself were detected along with the formation of C₂F₆(GWP = 12,400), a high GWP gas^41,43. In the case of C₂H₂F₄, more HF and COF₂ with low GWP values were formed compared to CHF₃, while the formation of CHF₃ having a high GWP decreased. Although CF₄(GWP = 7,380) was produced, it was found to be relatively small^44,45. For C₂H₂F₄+CF₄O, the formation of C₂F₆ was also observed compared to C₂H₂F₄, and the amount of COF₂ increased. The concentrations of HF and CHF₃ remained similar. However, the addition of CF₄O led to an increase in CF₄ formation during its dissociation and recombination. Fig 7(b) shows the MMTCE graph calculated when etching a ~50 nm high SiN_x target. When C₂H₂F₄-based gas was used instead of the conventional CHF₃-based gas, ~83.9% reduction of MMTCE was observed. In the case of using C₂H₂F₄+CF₄O-based gas, CF₄ was produced due to CF₄O, resulting in a ~75.2% reduction compared to the conventional CHF₃-based gas. However, the C₂H₂F₄+CF₄O -based etch process not only showed higher SiN_x etch rates and etch selectivity of SiN_x over SiO_x but also emitted much lower greenhouse emissions compared to the conventional CHF₃ process, confirming that it is an etching process that can replace the etching process using CHF₃, which is suitable for next-generation applications and is environmentally friendly.

Fig. 7

(a) Concentration of exhaust gases (ppm) measured using FT-IR in the exhaust line during plasma generation using CHF₃, C₂H₂F₄, C₂H₂F₄+CF₄O-based in Fig 3. (b) MMTCE values calculated for the gases used to etch the same SiN_x thickness.

Conclusions

This study investigated etch properties using low global warming potential (GWP) gases as an alternative to CHF₃, which has traditionally been used in SiN_x etch processes selective to SiO_x such as double patterning processes for FinFET, 3D NAND fabrication, etc. Such processes are directly relevant to advanced memory fabrication, where SiN_x is commonly used as a spacer or dielectric layer. These structures require precise profile control, minimization of polymer-induced defects, and compatibility with high-aspect-ratio integration. Replacing CHF₃ with C₂H₂F₄ resulted in higher SiN_x etch rates and higher etch selectivity over SiO_x, but trenching phenomena were observed in the etched pattern edge. By replacing it with C₂H₂F₄+CF₄O, not only higher SiN_x etch rate but also trenching could be eliminated.Plasma and surface analysis showed that, in the case of etch processing using CHF₃, the SiN_x etch rate was slow due to oxidation of the SiN_x surface due to the large amount of oxygen in the plasma, while in the case of C₂H₂F₄, the trenching phenomenon was observed by forming a polymer on the pattern surface. By adding CF₄O to the C₂H₂F₄, the polymer formed on the pattern surface could be successfully removed, and not only the SiN_x etch rate was increased, but also the trenching phenomenon was eliminated. In addition, the calculation of MMTCE showed that, when C₂H₂F₄ and C₂H₂F₄+CF₄O were used instead of conventional CHF₃, the MMTCEs were significantly reduced by ~83.9% and ~75.2%, respectively. These improvements, combined with significantly reduced MMTCE values, suggest that C₂H₂F₄+CF₄O is a promising candidate for eco-friendly and high-precision SiN_x etch processes applicable to advanced FinFET and 3D NAND device fabrication.