Hydrogen Peroxide Gas on the road from R&D to HVM for superior HZO films

Device shrinkage, three-dimensional and High Aspect Ratio (HAR) structures, and lower thermal budgets drive the continued search for new materials. A by-product of this search is a need for better oxidants for atomic layer deposition (ALD) and other thin film deposition processes.

While metal usage is rapidly expanding across the periodic table, oxidant choices are few: water, O2, ozone, and oxygen plasma being the leading choices for thin-film processing.

Each oxidant has its strengths and weaknesses. Plasma has limitations with the line of sight and may damage underlying sensitive channel materials or metal interconnects. Ozone is too aggressive with most metals. Water and oxygen are not reactive enough for today’s lower thermal budgets and more demanding precursors. Therefore, new oxidants could help address low-temperature thermal applications and simplify precursor design and selection.

At RASIRC, the investigation began for alternative oxidants when water vapor proved too limited for many ALD applications. Interest in delivering gas generated from hydrogen peroxide liquid began in 2007, with the first commercial sales in 2011 and 2012.

While the perception of the semiconductor industry is one of rapid innovation, the adoption of new technology is a slow process. If successful, it can frequently exceed a decade to reach high volume manufacturing.

Recently, RASIRC presented (April 2022 CMC2022, AZ, USA) benchmarking hydrogen peroxide vs. water and ozone in ALD of ferroelectric hafnium zirconium oxide (HZO). HZO is one of the primary candidate materials for new non-volatile memory using a capacitor device; it can be integrated into both Logic devices and as a stand-alone memory chip similar to Flash memory.

RASIRC and UT Dallas fabricated capacitor structures (MIM) and deposited HZO using water, ozone, or hydrogen peroxide at comparable process conditions.

The first finding was that the growth rate per cycle (GPC, below left) was considerably higher in the hydrogen peroxide case, essentially lowering the overall process time and precursor consumption of rather expensive Hafnium and Zirconium precursors. The hydrogen peroxide HZO films also proved to have a higher density (XRR, below middle) and lower etch rate (wet etch rate below right).

Growth rate per cycles, density by X-ray Reflectivity (XRR) and wet etch rate determination of HZO films deposited by ALD using either hydrogen peroxide, water or ozone.

Higher density metal oxide films are a sure sign of better electrical performance regarding high-k dielectrics and ferroelectrics. First, the hydrogen peroxide films showed a comparably higher effective k-value, lower leakage current (Jg), and could withstand a higher breakdown voltage (VBD), as seen below right. Water results were inferior to both ozone and hydrogen peroxide are not shown for clarity.

Indicative for ferroelectric phase content is a peak at approx. 2T= 30.3 deg and 35.8 (below right). In X-ray diffractograms, when comparing hydrogen peroxide vs. ozone, it[JS1] was shown that the hydrogen peroxide films could show a higher orthorhombic (ferroelectric) phase content at a lower thermal budget, i.e., the onset temperature for crystallization. Even though the orthorhombic ferroelectric phase is metastable over preferred tetragonal and monoclinic HfO2 and ZrO2 most stable phases, this can be understood that the atoms in higher density and purer hydrogen peroxide films will find their optimum positions under given conditions in the lattice faster due to less disturbance from contamination species that has to diffuse out of the lattice before a ferroelectric phase content can crystalize quenching the HZO films into the metastable ferroelectric phase.

Leakage (Jg) vs breakdown voltage (VBD) and gracing incidence x-ray diffraction (GI-XRD) after post deposition anneals for hydrogen peroxide HZO films compared to ozone HZO films.

The promising results above for higher quality ferroelectric films were then proven by complete ferroelectric electric characterization sweeping the current and voltage across the capacitor structures accordingly. As seen below in the P-E hysteresis curves, a clearly defined hysteresis response curve could be verified for hydrogen peroxide HZO films at a lower RTA temperature than for ozone films, the onset of 325 vs. 350 deg C. It may seem like a slight difference, but please keep in mind that the overall thermal budget for device integration in copper interconnect layers is in the range 350 to 390 deg.C depending on layer and technology node, and it is critical to stay below this temperature and as can be seen below this study yielded beautiful ferroelectric hysteresis at 350 deg. C for hydrogen peroxide, whereas ozone films had to go up to the danger zone of 400 deg. To do the same.

P-E hysteresis curves for hydrogen peroxide (right) and ozone (left) ferroelectric HZO films for different RTA thermal budgets.

Finally, TEM analysis showed that films could be downscaled to 5 nm film thickness and most probably below, staying perfectly intact even though a high roughness metal bottom electrode was used. 

High-resolution transmission electron microscopy (HR-TEM) of ferroelectric HZO films deposited by ALD and using RASIRC hydrogen peroxide technology.

To conclude, HZO ferroelectric films showed many advantages when hydrogen peroxide was employed compared to water and ozone:

· Higher device yield as measured in the number of functional ferroelectric capacitors

· Higher density films with lower wet etch rate

· Higher effective k-value

· Faster growth (ALD GPC)

· Lower film thickness for yielding films in electrical testing

· Lower leakage current and higher breakdown voltage

· Crystallization onset for ferroelectric phase content for lower thermal budgets (RTA temperature)

Next you can meet RASIRC at the AVS 22nd International Conference on Atomic Layer Deposition (ALD 2022), will be a three-day meeting 26-29th of July in Ghent Belgium, dedicated to the science and technology of atomic layer-controlled deposition of thin films and now topics related to atomic layer etching. Jeff Spiegelman, CEO of RASIRC will be presenting “Higher Effective Dielectric Constant of Hafnium Oxide When Grown with Hydrogen Peroxide Compared to Water Vapor” in session AF-MoP18 on 27th of June.

About RASIRC

RASIRC transforms liquids into dynamic gases that power process innovation in semiconductor and adjacent markets. By commercializing molecules for lower temperature processes, RASIRC patented technology enables the manufacture of atomic-scale oxides, nitrides, and metals. Innovative products such as BRUTE Peroxide, BRUTE Hydrazine, the Peroxidizer®, and Rainmaker® Humidification Systems are being used to develop solutions for 5G, AI, IOT, and advanced automation.

What makes RASIRC a unique industry leader is our technical expertise and commitment to solving complex industry challenges for our customers. Our team of industry experts has a proven track record of being first to market by efficiently delivering state of the art technology that reduces cost, improves quality, and dramatically improves safety. With our customers at the forefront of all we do, we continue to research, develop, and design innovative products that purify and deliver ultra-pure gas from liquids for the semiconductor and related markets. Contact RASIRC to help solve your complex problems. P: 858-259-1220, email info@rasirc.om or visit http://www.rasirc.com

RASIRC Granted Patent for Controlled Delivery of Hydrogen Peroxide Gas

Novel method for generation of H2O2 gas granted patent in US and Japan

San Diego, California – February 24, 2022 – RASIRC announced that the United States Patent and Trademark Office has granted Patent # US 11,154,792 B2 for a novel Method, System and Device for Delivery of Process Gas. The patent is applicable to RASIRC hydrogen peroxide products including the Peroxidizer®. The patent was also certified by the Japan Patent Office as Patent 6951321. The method enables more accurate and repeatable delivery of hydrogen peroxide gas into a wide range of flow rates, operating pressures, and temperatures.

This invention correlates instantaneous applied power to mass delivery of H2O2, largely eliminating nonuniformities in the liquid source and thermal droop. This enables the Peroxidizer to provide accurate and linear delivery of chemistry without regard to vaporization temperature or process pressure.

“By applying this new process control method, the Peroxidizer can provide linear control throughout the mass flow range independent of the carrier gas flow rate and operating pressure of the process,” said RASIRC Founder and CEO Jeffrey Spiegelman. “This design integrates custom hardware, firmware and software to improve both Peroxidizer performance and tool-to-tool repeatability, helping us meet the semiconductor standards for high volume manufacturing (HVM).”

Power control is a very effective way to deliver vapor from a liquid source. Most liquid to vapor mass flow control is managed by temperature regulation of the liquid source or bath. However, bath temperature increases with increased mass flow rate causing temperature non-uniformities, localized droop, and output instabilities. As the temperature increases the vapor pressure increases on a power curve. Error increases in a highly non-linear fashion. For example, H2O2 vapor pressure change of 1°C at 90°C is 20 times larger than 1°C at 30°C, leading to 20X increased error at the higher temperature. Power control is based on the mass evaporated so the error does not increase with increasing bath temperature.

About the RASIRC Peroxidizer®

The RASIRC Peroxidizer provides a safe, reliable way to deliver high-concentration hydrogen peroxide gas into ALD, annealing, gapfill, dry surface preparation, and cleaning processes.

About RASIRC

RASIRC transforms liquids into dynamic gases that power process innovation in semiconductor and adjacent markets. By commercializing molecules for lower temperature processes, RASIRC patented technology enables the manufacture of atomic-scale oxides, nitrides, and metals. Innovative products such as Brute Peroxide, Brute Hydrazine, the Peroxidizer, and the Rainmaker Humidification Systems are being used to develop solutions for 5G, AI, IOT, and advanced automation.

What makes RASIRC a unique industry leader is our technical expertise and commitment to solving complex industry challenges for our customers. Our team of industry experts has a proven track record of beating larger competitors to market by efficiently delivering state of the art technology that reduces cost, improves quality, and dramatically improves safety. With our customers at the forefront of all we do, we continue to research, develop, and design innovative products that purify and deliver ultra-pure gas from liquids for the semiconductor and related markets. Contact RASIRC to help solve your complex problems.

P: 858-259-1220, email info@rasirc.om or visit http://www.rasirc.com

The Emergence of Hydrazine (N2H4) in Semiconductor Applications

 

The Emergence of Hydrazine (N₂H₄) in Semiconductor Applications

by Jeffrey Spiegelman and Daniel Alvarez

 Purpose

Historically, metal-nitride MOCVD and ALD films have been fabricated with Ammonia (NH₃).  However, lower thermal budgets and shrinking 3-dimensional structures are needed for next generation semiconductor devices.  These challenges have exposed limitations with ammonia which could be overcome by replacing ammonia with hydrazine (N₂H₄).  Purity of commercially available hydrazine has  delayed its adoption.  RASIRC Inc. has recently developed a new formulation of hydrazine called BRUTE® Hydrazine which is safer and meets purity requirements for semiconductor manufacturing. Prior to Brute Hydrazine, the body of technical data applicable to semiconductor processing was limited and scattered.  This paper provides an overview of the growing activity in the thin film use of Brute hydrazine as well as early references on laboratory grade hydrazine for historical completeness.

Increasing Need for More Reactive Nitrogen Sources

Emerging devices such Logic and Advanced Memory require high quality thin (5-20 Å) electrode and barrier films.  Difficult thermal budget constraints are now being placed on well-known materials such as SiN_x, TiN_x and TaN_x.^1-3  Deposition temperature limitations have dropped to 350°C and below while very low resistivity (<150 micro-ohm/cm) for TiN and TaN must still be achieved.   Although metal and nitride films grown using plasma assisted processes (PE-ALD) and (PE-CVD) at low temperatures exhibit enhanced properties, the damage induced by plasma on sensitive substrates is one of the common drawbacks,^4,5 as well as inability to support HAR or three-dimensional structures like horizontal vias and deep trenches.

III-V Nitride devices require a more reactive nitrogen source to reduce deposition temperatures and increase compositional stability.⁶ Growth rates for InGaN films deposited with ammonia at reduced temperatures are prohibitively slow and grossly inefficient in ammonia usage. A more reactive nitrogen source can enable acceptable deposition rates at 500-650° C, where alloy stability is significantly increased and nitride source to precursor ratio can be reduced.

In addition to growing thin nitride films, hydrazine can also act as a reducing agent for several late transition-metals.  This work is highly relevant to the use of hydrazine as a surface cleaning agent as well as a potential adder for metal ALD.⁷

Figure 1: Low Temperature Thermal ALD growth rate with Hydrazine comparable to PEALD with Ammonia.

The following table provides primary references for the areas of hydrazine ALD/CVD relevant to Semiconductor device applications.  Additional relevant references on related films are also included.

 

Precursor and Temperature	Film	Reference
Al surface nitridation 200C-450C	AlN	Taylor et.al.  U.S. Patent 6465350, 2002
TMA MOCVD  300C-400C	AlN	Fujieda, S. et. al.   Adv. Func. Mat.  1996, 6(3), 127-134
TDEAA  150C-225C	AlN	Abdulagatov, A.I.  et. al. Russian Microelectronics, 2018, 47(2), 118–130.
TMA 175C-350C	AlN	Jung, Y.C.  et. al. Materials 2020, 13, 3387;  https://doi:10.3390/ma13153387
TDMAA 225C-400C	AlN	Ueda, S.T. et. al.  Appl. Surf. Sci.  2021, 554, 149656
BCl3 ,  350C	BN	Wolf, S. et. al. Appl. Surf.  Sci.   2018, 439, 689–696
Surface Clean 200C	Cu	Hwang, S.M.  et. al. ECS Trans. 2019, 92, 265
Surface Clean 100C-300C	Cu, Co	Hwang, S.M. “Effect of Surface Cleaning Efficacy on Vapor-Phase Cleaning of Cu and Co Using Anhydrous N2H4” AVS ALD/ALE 2021 Session: Area Selective ALD AS4-1
TMG, 400C-800C	GaN	Fujieda, S.  et. al. Jpn. J. Appl. Phys.  1987, 26, 2067-2071
TMG, TMI, 600C-900C Theoretical	GaN, InGaN	Koukitu, A. et. al. phys. Stat. sol. (b), 1999, 216(1), 707-712
TMG Theoretical	GaN	Goddard, W. et.al. J. Phys. Chem. C 2015, 119(8) 4095–4103
[Ru(DMBD)(CO)3] 200C. Metal Deposition	Ru	Cwik, S. et. al.  J. Vac. Soc.  Sci. & Tech. A 2020, 38, 012402; https://doi.org/10.1116/1.5125109
SiH4   550C-1050C	SiN	Yoshioka, S. et. al.   J. Electrochem. Soc. 1967, 114, 962–964.
SiH4/W hot wire  300C	SiN	Matsumura, H. 1989 Jpn. J. Appl. Phys. 28 2157
Si2H6, Si3H8  350C-550C	SiN	Kanoh, H.  et al.  “Low-Temperature Chemical-Vapor-Deposition of Silicon Nitride” Journal de Physique IV Proceedings, 1991, 02 (C2), pp.C2-831-C2-837.
Si surface Nitridation. 300C-500C	SiN	Abyss, J.A. et. al.  J. AIChE  1995, 41, 2282–2291
Si2Cl6  285C	SiN	Edmonds, M. et. al.,   J. Chem. Phys. 2017, 146, 052820 ;  https://doi.org/10.1063/1.4975081
Si2Cl6  320C-410C	SiN	Kondusamy, A.  et. al. “Low Temperature Thermal ALD of Silicon Nitride Utilizing a Novel High Purity Hydrazine Source”, Electrochem. Soc. AiMES 2018, Meet. Abstr.  G02-993
Si2Cl6  410C-650C	SiN	Le, D.N. et al “Thermal Atomic Layer Deposition of Silicon Nitride Using Anhydrous Hydrazine and Ammonia” AVS ALD 2021, Session AF9.
TBTDET 150C-250C	TaN	Burton, B.B., et. al. J. Electrochem. Soc. 2008, 155, D508
TBTDET 100C-300C	TaN	Wolf, S. et.al.  Appl. Surf. Science, 2018, 462, 1029-1035
TDMAT  200C	TiN	Wierda, D.A. et. al. Electrochemical and Solid-State Letters, 1999, 2 (12) 613-615
TiCl4  200C-350C	TiN	Abdulagatov, A.I.  Ph.D. Thesis, Univ. of Colorado, 2012, UMI No. 3549153
TiCl4  300C-400C	TiN	Wolf, S. et.al.  Appl. Surf. Science, 2018, 462, 1029-1035
TiCl4  300C-400C	TiN	Kuo, C.H. et. al. “Low Resistivity Titanium Nitride Thin Film Fabricated by Atomic Layer Deposition on Silicon”   AVS ALD 2021, Session AM5-9.
TiCl4  250C-400C	TiN	Alvarez, D. et. al.  “Comparative Study of Titanium Nitride ALD using High Purity Hydrazine vs Ammonia” Electrochem. Soc. 2020 Meet. Abstr. MA2020-02 1668
BTBMW 300C	WN	Bernal-Ramos, K.   Ph.D. Thesis, Univ. of Texas, Dallas, 2014, UMI  No. 3668896
BTBMW 250C-350C	WN	Le, D.N.  et.al. “Atomic Layer Deposition of Nanometer Thick Tungsten Nitride Using Anhydrous Hydrazine for Potential X-Ray Optics Application” AVS ALD/ALE 2021 Session: AF10-15

Discussion on Specific Films

Titanium Nitride (TiN) is a critical film in semiconductor manufacturing. Commonly TiN is utilized as an electrode material as well as a low resistivity barrier layer.  Early CVD work by Wierda demonstrated low temperature (50C-250C) TiN CVD by hydrazine and TDMAT.  Optimal results were obtained when 1.9% hydrazine was combined with ammonia.  This may be attributed to a different mechanistic pathway or ammonia dilution of oxygen containing contaminants.  Wolf later demonstrated low temperature (300C) TiN ALD with the use of TiCl₄. This result was then optimized by Kuo in the same lab, where resistivities well below 180 micro-ohm/cm were achieved by reducing oxygen contamination in the film through improved hydrazine purity. A comparative study of Hydrazine vs Ammonia for TiCl₄ was reported by Taiyo Nippon Sanso, where the two nitrogen sources showed highly disparate growth rates and film properties.  Hydrazine demonstrated viability at the 250C-400C range for low temperature semiconductor applications.

Silicon Nitride (SiN) is a widely used material in semiconductor devices. SiN is commonly used as an etch stop, a dielectric layer, an encapsulation layer, and as a barrier layer on organic devices.  As early as 1967, hydrazine and Silane CVD was demonstrated at 550C.  This work was then followed-up by Kanoh with higher silanes in the 350C-550C range.  In a very interesting approach, Abyss demonstrated Si surface nitridation with hydrazine at temperatures as low as 300C.  More recently, Edmonds cleverly used hydrazine/hexachlorodisilane ALD to place a thin SiN passivation layer on SiGe at 285C.  Extensive studies have been carried out by the Kim group at UT Dallas in the range of 320C-650C.  Below 400C, thermal ALD leads to films with good composition, but unfavorable low density and high wet etch rates.  This can be overcome with addition of Argon plasma densification.  At 480C and above, thermal ALD films are grown with high density, low wet etch rates, and reduced hydrogen incorporation.  When compared to ammonia grown films in the same temperature range, the hydrazine ALD films are superior up to temperatures >600C where films properties become more similar.

Gallium Nitride and Indium Gallium Nitride (GaN, InGaN) grown with hydrazine have had few publications in the last 20 years despite interest in reduction of ammonia usage and poor indium incorporation. These films are central in LEDs and emerging power devices.  Fujieda demonstrated that overall chemical consumption can be greatly reduced with hydrazine vs ammonia for GaN deposition in the 400C-800C range.  Koukitu followed this up with a theoretical thermodynamic study showing how the use of hydrazine can reduce deposition temperature and stabilize composition for GaN and InGaN films.  In 2015, Goddard elucidated the likely mechanisms for hydrazine vs ammonia is GaN deposition.

Though little has been published for GaN/InGaN deposition with hydrazine, viability for III/V materials can be inferred from work published for AlN ALD with hydrazine.  Fujieda reported MOCVD with trimethyl aluminum (TMA) in the 300C-400C range.  More recently Jung reported ALD with TMA as low as 175C and compared to ammonia in the 175C-350C range.  Abdulagatov made use of the nitride-based ligands with TDEAA/hydrazine ALD in the 150C-250C range.  In a similar approach using TDMAA, Ueda has reported the deposition of crystalline AlN films as low as 350C with thermal ALD.  With the addition of Argon plasma densification, crystalline films can be obtained as low as 225C, where crystallinity in AlN was optimized at 400C.

Copper, Cobalt and Ruthenium can be reduced in situ by Hydrazine.  Furst provided a detailed review on hydrazine as a reducing agent for organic compounds.⁸ Recently Hwang reported an extension of this reactivity to Cu surfaces.  Gas phase reduction of Cu oxides to Cu metal with hydrazine at moderate temperatures (100C-300C) was reported.  Here, hydrazine is introduced in short pulses, analogous to an ALD reaction. A similar report for Cobalt has also been presented by Hwang.  Cwik working in the Winter group has recently released data showing the ability to grow Ru metal using hydrazine as a reducing agent in Ru ALD at 200C.  Here hydrazine was found to be advantageous over substituted hydrazine derivatives.

Conclusion

Hydrazine is emerging as a replacement for ammonia in low temperature applications.   Recent examples of different production-worthy nitrides have been reported for both ALD and MOCVD films.  These positive reports have led to an increasing level of interest within the scientific community looking for solutions to new device structures and increased density.

Contact the Author

The author is available for additional technical discussion. Contact RASIRC to schedule an appointment.

References

1. Burton BB, Lavoie AR, George SM (2008) Tantalum nitride atomic layer deposition using (tert-Butylimido) tris(diethylamido)tantalum and Hydrazine. J Electrochem Soc 155, D508

2. Alvarez, D.; Spiegelman, J.; Andachi, K.; Holmes, R.; Raynor, M.; and Shimizu, H. Enabling Low Temperature Metal Nitride ALD Using Ultra-High Purity Hydrazine: ET/ID: Enabling Technologies and Innovative Devices. 2017 28th Annu. SEMI Adv. Semicond. Manuf. Conf., Saratoga Spring, NY, USA, 2017, 426–430.

3. Hwang, S.M.; Kim, H.S.; Le, D.N.; Ravichandran, A.V.; Sahota, A.; Lee. J.; Jung, Y.C.; Kim, S.J.; Ahn, J.; Hwang, B.K.; Lee, L.; Zhou, X.; and Kim, J. Plasma-Enhanced Atomic Layer Deposition of Nanometer-Thick SiNx Films Using Trichlorodisilane for Etch-Resistant Coating. ACS Appl. Nano Mater. 2021, 4, 2558–2564. https://doi.org/10.1021/acsanm.0c03203.

4. Kim, H.; Oh, I.-K.; Review of Plasma-Enhanced Atomic Layer Deposition: Technical Enabler of Nanoscale Device Fabrication. Jpn. J. Appl. Phys. 2014, 53, 03DA01. https://doi.org/10.7567/JJAP.53.03DA01.

5. Mussroot, J. et.al. Microelectronic Engineering 86 (2009) 72-77. http://dx.doi.org/10.1016/j.mee.2008.09.036

6. Ravinder Kour et al 2020 ECS J. Solid State Sci. Technol. 9, 015011

7. Hwang, S. M.; Peña, L. F.; Tan, K.; Kim, H. S.; Kondusamy, A. L. N.; Qin, Z.; Jung, Y. C.; Veyan, J.-F.; Alvarez, D.; Spiegelman, J.; et al. Vapor-Phase Surface Cleaning of Electroplated Cu Films Using Anhydrous N2H4. ECS Trans. 2019, 92, 265–271.

8. Furst, A. et. al. Chem. Rev. 1965, 65, 51–68.