Thin films for 3D: ALD for non-planar topographies

06/01/2009

Extreme and non-planar substrate topographies that provide challenges for conventional thin film deposition techniques are abundant. Such applications span the range of structures from DRAM trench cells and MEMS devices to coatings of esoteric materials such as aerogels, viruses, nanotubes, microlenses, piping, grains and particles, and even porous dental material. For these types of important and interesting applications, atomic layer deposition (ALD) provides an exceptional technology for overcoming such challenges in order to achieve superior film uniformity, step coverage, and film conformality.

Ganesh M. Sundaram, Eric W. Deguns, Ritwik Bhatia, Mark J. Dalberth, Mark J. Sowa, Jill S. Becker; Cambridge NanoTech Inc., 68 Rogers St. Cambridge, MA USA

Following the seminal work of Aleskovskii [1] and Suntola [2] in establishing the field of ALD science, the subject has rapidly expanded in both academic and commercial arenas to encompass a wide variety of applications. The proliferation of this technology has been principally due to the advantages offered by the deposited film quality and conformality, resulting from the inherent growth mechanism found in ALD. Films are deposited by sequentially pulsing appropriate precursor material into the ALD reaction chamber, followed by a purge cycle of an inert gas. This sequence is repeated again for each subsequent layer. A key element of the deposition is the self-limiting nature of the process that allows repeatable monolayer-by-monolayer growth, with a fairly broad process window. The resulting films are pinhole-free, uniform, and extremely conformal.

ALD growth mechanism

The ALD material system that exhibits these characteristics best is Al₂O₃. It is typically grown thermally at 80?????250??C with trimethylaluminium (TMA) and water as the precursors. The reactions are as follows for growth on a Si surface [3]: Volatile TMA reacts in a self-limiting fashion with hydroxyl groups terminating the Si surface. This reaction deposits aluminum atoms with dangling methyl groups, while excess TMA and methane are pumped away. Next, water is introduced, which readily reacts with the methyl groups resulting in the formation of Al-O bridges and the regeneration of surface hydroxyls. This serves to ready the surface for the next cycle, starting with another TMA pulse. Defect-free layers of Al₂O₃ are formed in each complete cycle with a growth rate of ~1.0Å/cycle achievable in a large part of the process window.

The reaction mechanism for other materials, such as nitrides and sulfides is similar, given the reaction of the second precursor with the chemisorbed surface species. However, the ALD of certain metals (i.e., platinum or ruthenium films) proceeds primarily though a combustion mechanism. In such cases, the second reactant is usually oxygen or ozone, which helps convert the organic ligands of the precursor into volatile carbonates (CO and CO₂) and the reduction of the precursor into a metallic state.

ALD and high-aspect ratio topography

Along with the superior flat film uniformity that can be achieved through ALD, a striking feature of the deposition process is the ability to conformally coat 3D geometries and high-aspect ratio (HAR) features. To take full advantage of this feature, an ALD system is required to have a true closed-vacuum reaction chamber. This provides the user with the needed flexibility to moderate the exposure time of the samples to the reactants. In this manner, small precursor pulses can be held in the reaction chamber for longer periods to increase the exposure of the sample to the growth chemistries. This is the key to depositing conformal coatings on structures with extreme geometries and aspect ratios.

An elegant and simple mathematical model that explains the conditions required to coat HAR holes, trenches, and arbitrary shapes has been described in detail by Gordon et al. [4, 5]. The results of the model show that for a hole of length l, diameter d, and aspect ratio a, the exposure (Pt) required for coating the sidewalls of the hole is given by:

Click here to enlarge image

Where P = reactant partial pressure, and t = time needed to coat the entire hole length
T= temperature
k = Boltzmann constant
m = molecular mass of the reactant
S = saturated surface coverage per cycle (no. of molecules/m2)

Coating of the bottom of the hole is given by the following expression:

Click here to enlarge image

Combining equations 1 and 2 gives the total exposure needed to cover both the sidewalls and bottom of the hole.

Click here to enlarge image

It is evident from the model that as the aspect ratio of the features to be coated increases, the exposures increase, and for ultra-high aspect ratios, the exposure scales as the square of the aspect ratio -(Pt) α a². Note that the Gordon model has been further refined by the Detavernier group at the University of Ghent [6]. The results of the model are consistent with those of Gordon, and do not change the essential truth — coating extreme topographies can absolutely be accomplished, but the processes require time.

Applications

DRAM trench structures. The DRAM trench capacitor is a classic example of an HAR feature where ALD is consistently the method of choice for coatings. The requirement for ever-increasing levels of charge storage capability has in turn driven the aspect ratios of these structures to more extreme values. Figure 1 is an example of an ALD deposition of SiO₂ using a silanol process — t-butoxy silanol and TMA in an exposure mode of growth [7]. The trench structure consists of a 100nm opening, and is 7µm long. The aspect ratio coated is 70:1. The film is uniform and conformal, as measured through a cross-section of the device.

Figure 1. Cross-sections of holes 7µm deep and 100nm in diameter. On the left is a complete uncoated hole. On the right are higher magnification images of the top, middle, and bottom of a hole coated conformally with a uniform silica film 46nm thick made by four ALD cycles.Click here to enlarge image

null

Figure 2. Deposition of tungsten nitride through a fused silica capillary. The capillary has a 20µm opening. The deposition extends 4.2µm into the capillary, starting at the left arrow and ending at the right.Click here to enlarge image

While sophisticated methods can be used to determine the aspect ratio infiltrated by the coating, some simpler methods can be used to evaluate to what extent a coating infiltrates a feature. One such method, pioneered by Becker, is the use of thin capillary samples [8]. Thin capillary tubes can be placed in the chamber and deposited with an ALD coating. Following the coating, the samples are removed and the outer coating of the capillary tube is removed by heat. Next, refractive index matching liquid is allowed to be drawn into the capillary. The tube is then placed under a microscope for examination. Under examination, the portion of the capillary interior that is uncoated disappears from view due to the presence of the index matching fluid, while the portion of the capillary that is coated appears clearly. An example of this is shown in Fig. 2. The image shows a fused silica capillary with an opening of 20µm, where a deposition of tungsten nitride (WN) has traveled 4.2mm into the structure. This provides a simple method to determine the extent of the infiltration of the thin films into the sample structure.

Aerogels. Another example of an application with strong potential for industrial uses is the coating of aerogels. Aerogels are extremely low density materials initially produced by drying gels in such a manner as to remove their liquid content, while leaving the solid matrix of material behind [9]. This results in extraordinarily porous material with a dendrite-like supporting structure, whose properties lend themselves very well to insulator, desiccant, and load-bearing applications. Silica is the most common type of aerogel. ALD can be used to infiltrate the pores of the aerogels creating a method for developing extremely high surface area materials—a key element of catalytic applications.

Figure 3. SEM image of a cross-section of an aerogel. EDS analysis of the silica aerogel, shows ~100µm penetration of the Al2O3 film into the structure. Pore size is 50nm. Aspect ratio is 2000:1.Click here to enlarge image

An example of an aerogel coated using Al₂O₃ is shown in Fig. 3. The sample consists of a silica aerogel with a pore size of 50nm. The Cambridge Nanotech Savannah ALD system’s exposure mode was used to deposit an Al₂O₃ film into the aerogel. Exposure times on the order of 180 secs were used.

Nanostructure formation. The use of ALD can also be extended to the formation of extremely interesting nanostructures which are fabricated through additive and subtractive processes. Leaders in this field have been the Nielsch and Knez groups at the Max Planck Institute in Halle, Germany (MPI-Halle); though as this article goes to press, Nielsch is now at the University of Hamburg. One example involves the creation of nanotubes of ZnAl2O4 [10]. The fabrication begins with a nanowire of ZnO. On this structure Al₂O₃ is conformally coated using ALD. Following the deposition, the ZnO-Al₂O₄ structure is annealed. The annealing process eliminates the ZnO core nanowire, and leaves behind a nanotube of ZnAl₂O₄.

Figure 4. Micrograph of tobacco mosaic virus. ALD-based TiO2 was deposited to create a template for 3D nanostructures.Click here to enlarge image

Another example (Fig. 4) involves the use of biological materials as a template for creating the structure [10]. Here, the tobacco mosaic virus is used for that function. TiO₂ was deposited using Exposure Mode in a Savannah S100 system. The virus was of length 300nm, OD 18nm, ID 4nm. ALD work on such types of macromolecules opens the doors for using simple biological material as templates for the fabrication of 3D semiconducting/metallic/insulating nanostructures, with minimal reliance on standard lithographic patterning.

Macro size features

The issues of extreme topography are not limited to nanosize devices. Often, conceptually simple requirements for large objects, such as coating the interior of a long pipe with a convoluted interior, can generate the same challenges — namely the ability to achieve a uniform coating when large aspect ratios are involved. Again, this can be handled using ALD in exposure mode. Other examples of macro-sized features include arrays of micro drill bits that require coatings of thin film with anti-wear material. The geometry of the bits consists of complex curves, and shadowed surfaces, which would be difficult to coat by conventional techniques but can be coated in a batch process by ALD.

Particles and powders

Particles and powders represent a class of non-planar samples, where the potential for useful applications is large provided they are able to be coated. For example, alumina coatings of boron nitride (BN) particles help to reduce their thermal conductivity and improve their surface wetting characteristics, making the coated particles an excellent filler material for microelectronic packaging. Applications similar to this also exist in the biological and consumer fields.

The main problem with coating particles stems from their irregular shapes, which include undercut profiles, additionally, the act of coating them could lead to aggregation (clumping together) of the particles. There are ALD methods, many of them developed by the George Group at the University of Colorado at Boulder, to overcome these issues [11]. The main feature of these designs is to maintain sufficient flow/movement around the particles during the ALD growth process, such that the particles can remain separated from one another. The ALD growth in turn allows the irregular sizes and shapes of the particles to be coated uniformly over the entire surface.

Conclusion

While planar surfaces continue to require thin film coatings, 3D devices and objects with stringent coating requirements continue to emerge. These application spaces are greatly aided by the use of ALD technology, which serves to fulfill the requirements and illuminate the path in the exploration for ever-increasing thin films coating for extreme geometries.

References

1. A.M. Shevjakov, G.N. Kuznetsova, and V.B. Aleskovskii, in Chemistry of High Temperature Materials, Proceedings of the Second USSR Conference on High-Temperature Chemistry of Oxides, Leningrad, USSR, 26-29, 1965 (Nauka, Leningrad, 1967), pp. 149-155, in Russian.
2. T. Suntola, and J. Antson, US Patent No. 4058430 (15 Novenber 1977).
3. R.L Puurunen, J. Appl. Phys. 97, 121301, (2005), also www/cambridgenanotech.com.
4. Roy G. Gordon, Dennis Hausmann, Esther Kim, and Joseph Shepard, Chem. Vap. Deposition, 9, (2), 73, (2003).
5. S.O. Kucheyev, J. Biener, T.F. Baumann, Y.M. Wang, A.V. Hamza, Z. Li, D.K. Lee, and R.G. Gordon, Langmuir, 24, 943, (2008).
6. C. Detavernier, J. Dendooven, D. Deduytsche, and J. Musschoot, ECS Transactions, 16, (4), 239, (2008).
7. D. Hausmann, J. Becker, S. Wang, and R.G. Gordon, Science 298, 402 (2002).
8. J.S. Becker, Ph.D Thesis, (2003).
9. S. S. Kistler, J. Phys. Chem. 34, 52, 1932.
10. M. Knez, K. Nielsch, A.J. Patil, S. Mann, and U. Gösele, ECS Transactions 3, (15), 219 (2007).
11. H. J. Fan, M. Knez, R. Scholz, D. Hesse, K. Nielsch, M. Zacharias, and U. Gösele, Nano Letters 7, (4), 993 (2007).

Ganesh M. Sundaram received his PhD in physics at Oxford U. and is VP of Technology at Cambridge NanoTech Inc., 68 Rogers Street, Cambridge, MA, 02142 USA; ph.: 617-674-8800; [email protected].

Eric W. Deguns received his PhD in chemistry at the U. of California ??? Santa Barbara and is an ALD application scientist at Cambridge NanoTech Inc.

Ritwik Bhatia received his PhD in chemical engineering from Purdue U. and is a senior research scientist at Cambridge NanoTech Inc.

Mark J. Dalberth received his PhD in physics at the U. of Colorado, Boulder and is a physicist/principal engineer at Cambridge NanoTech Inc.

Mark J. Sowa received his PhD in chemical engineering from Princeton U. and is a senior research scientist at Cambridge NanoTech Inc.

Jill S. Becker received her PhD in chemistry at Harvard U. She founded Cambridge NanoTech in 2003.