ReviewMechanisms of surface reactions in thin solid film chemical deposition processes☆
Introduction
Thanks in great part to their isotropic nature, chemistry-based processes for solid film depositions have gained much interest in connection with several industrial applications in recent years. In the microelectronics industry in particular, they can be used to grow thin films conformally and to coat topologically complex surfaces in order to create conducting or insulating layers or to introduce diffusion, adhesion, or protection barriers. Standard chemical vapor deposition (CVD) methods have been available for these uses for some time already [1], [2], but recent interest has focused on a variation known as atomic layer deposition (ALD) where the overall chemical process is split into two or more self-limiting and complementary reactions in order to gain better control over the film growth at a submonolayer level (Fig. 1a) [3], [4]. ALD has been identified as one of the most promising technologies for the deposition of thin films in future modern microelectronics manufacturing [5], [6].
One advantage of CVD and ALD is that they offer many options in terms of the selection of the compounds and reactions to be used to deposit a particular material on surfaces. On the other hand, most of those options require complex chemistry, often including undesirable side reactions that may lead to the incorporation of impurities in the grown layer or to surface etching, and more generally to poor film quality (Fig. 1b). To avoid such complications, early chemical deposition processes were designed around simple precursors such as metal halides [7]. Unfortunately, many elements do not form such simple compounds; for the delivery of late transition metals, for instance, it is often necessary to use organometallic or metalorganic compounds. Another consideration when choosing CVD or ALD precursors is that they not only are required to follow clean surface conversion to the desired film material, but also need to be sufficiently volatile and stable so that they can be easily delivered intact to the deposition reactor. Much of the modern work in ALD is focused on designing, synthesizing, and testing viable new precursors that fulfill those requirements [8].
In spite of the limitations mentioned above, the field of ALD has blossomed in recent years. As the microelectronics industry advances toward the 14 nm node and beyond, the need for the controllable deposition of thin films conformally is becoming increasingly pressing, and ALD is developing as one of the few options available to tackle this issue. Moreover, the applications of ALD are being extended well beyond its original uses into other fields such as the manufacturing of optical and magnetic devices, flat panel displays, catalysts, protective coatings, and textiles, and also to areas related to the conversion (solar cells), utilization (fuel cells), and storage (batteries and supercapacitors) of energy [9], [10], [11]. Much work is being dedicated to the identification, testing, and assessing of the use of ALD for such practical applications. An increasing number of synthetic research groups are also developing new promising precursors for the delivery of virtually every element in the periodic table in these chemical-based deposition processes. It would be highly desirable to have a good understanding of the surface chemistry underpinning the ALD processes in order to design precursors in a rational way, from first principles.
Unfortunately, studies in this direction have lagged those on the more practical aspects of film deposition, although that is starting to change [12]. There have certainly been important contributions in this area worth mentioning here. For one, there have been a number of insightful reports on the characterization of the overall stoichiometry of the reactions that take place in each ALD half cycle by using quartz crystal microbalances (QCM), a technique used to follow the changes in mass that occur on the surface upon exposure to each of the reactants [13], [14]. It should be noted, however, that QCM measurements do not provide any chemically specific information. A complementary approach to QCM in these studies is the use of mass spectrometry for the detection of the gas products downstream from the ALD reactor [13]. Mass spectrometry is certainly a powerful technique, employed in alternative ways in temperature-programmed desorption, molecular beams, and GC/MS analysis in our laboratory, but it is sometimes difficult to differentiate among the several desorbing species that may be produced in ALD systems using complex metalorganic precursors. Also, some of the gas-phase species generated during ALD processes may not be stable and may undergo further conversion before detection. In terms of the characterization of surface intermediates, perhaps the most explored approach has been the use of infrared absorption spectroscopy (IR) [15], [16]; it is unfortunate that, to date, this technique has not lived up to its potential in ALD studies, perhaps because the experiments are not easy to carry out and the data not always straightforward to interpret. Overall, more mechanistic studies are still needed to develop a more complete picture of the mechanistic details of ALD reactions, possibly combining several techniques and approaches.
We in our laboratory have taken a modern surface-science approach to the study of the mechanism of the surface reactions associated with ALD processes, by using a combination of surface-sensitive analytical techniques such as X-ray photoelectron spectroscopy (XPS), temperature programmed desorption (TPD), low-energy ion scattering (LEIS), and IR to investigate the details of those reactions at a molecular level [12], [17], [18], [19], [20]. Our initial focus has been on applications of ALD for the back-end-of-line processing in integrated circuit fabrication, where the individual devices (transistors, capacitors, resistors, etc.) made on the silicon wafer are wired together [21]. The surface chemistry of a number of the copper precursors proposed in the literature for this use, including copper acetamidinates [22], [23], [24] and copper acetylacetonates [25], has been investigated. In addition, the surface processes involved in the formation of early transition-metal nitrides, titanium and tantalum nitrides in particular, using either halides [26], [27] or amido [27], [28], [29], [30], [31], [32] metal complexes, has been studied as well; such metal nitrides have been considered for the formation of diffusion barriers to prevent the electromigration of copper into the underlying silicon surfaces [33]. More recently, our diffusion-barrier work has been extended to the deposition of self-forming manganese-based films [34] using carbonyl [35], cyclopentadienyl [36], [37], [38], and acetamidinate [38] Mn complexes. We have also studied the surface chemistry of a series of metal carbonyls in connection with the deposition of films of several late transition metals [21], [39], [40], [41], [42], [43], [44], and within the last year have started to investigate the surface thermal conversion of strontium imidazolates as a way to grow high dielectric materials.
In this review, an overview is provided of the results that have derived so far from our studies on the mechanism of ALD processes. We draw from the specific surface chemistry identified in the systems mentioned above to extract general conclusions relevant to the design of film deposition processes (Fig. 1b). In terms of the thermal conversion of the reactants, of both the main metal precursors and the co-reactants, on the surface, we tackle issues having to do with reaction stoichiometry, redox chemistry involving the metal center, the role of the co-reactant, and the thermal conversion of adsorbed ligands. We also address the participation of the underlying surface in the promotion of the ALD reactions, in particular the activity of minority nucleation surface sites such as hydroxyl groups as promoters for the adsorption and conversion of the precursor, and also discuss the potential evolution of the grown films toward the formation of new complex multi-component structures. We close with some remarks on our personal view of the importance of developing a molecular-level understanding of the chemistry of ALD to the design of appropriate ALD precursors and ALD process conditions.
Section snippets
Reaction stoichiometry
The chemistry of CVD and ALD processes is so poorly understood that even their overall stoichiometry is often not known. Many ALD processes have been derived from similar pre-established CVD systems, but even in those cases, only limited experimental information is available on the chemical equation that accounts for the chemistry used [45]. In general, most researchers have focused on the ability to deposit the solid material of interest, and have relied on chemical intuition, thermodynamics
Changes in metal oxidation state
One of the things that complicate the stoichiometry of the ALD reactions designed to deposit titanium nitride films is the fact that such conversions involve the reduction of the titanium metal center. Most precursors available for the ALD of Group IVB metals, not only of titanium but also of zirconium and hafnium, are based on metal centers in a +4 oxidation state. This is the case with metal halides such as TiCl4, but also with popular amido ALD precursors such as TDMAT [62] and
Role of co-reactant
A key requirement in CVD and ALD processes is that they must involve chemistry that allows for the removal of all elements other than the ones being deposited. The simplest way to accomplish such a goal would be for the ligands in the main precursors to be removed from the surface cleanly by the second agent, the chemical used in the second half cycle in ALD. However, since most organic ligands used in metalorganic ALD precursors are not neutral, their removal may require additional chemistry.
Thermal decomposition of ligands on the surface
In addition to the chemistry that the adsorbed ALD precursors follow with the co-reactants, it is also necessary to consider the inherent thermal behavior of the precursor ligands themselves on the surface. This factor becomes increasingly more relevant as more complex organic ligands are introduced in the design of ALD precursors. Ideally, such ligands are to remain intact upon adsorption and to be removed by the co-reactant in the second half of the ALD cycle, as discussed before. Simple
Role of the nature of the surface on the chemistry of ALD
As mentioned above, the initial surface chemistry of ALD precursors depends on the nature of the substrate where the films are grown. On most semiconductors and insulating materials, this chemistry is quite localized, involving specific atoms on the surface. On metals, where the electronic density is more delocalized, the description of surface bonding may be somewhat more complex, but even there the relevant organic chemistry is likely to be molecular [17], [19], [104]. This means that ALD
Nature of the film grown by ALD
In the previous section, it was mentioned that both the nature and density of nucleation sites and the relative reactivity of ALD precursors on those sites versus on the rest of the substrate and on the new surfaces built by the film deposition can determine the structural details of the film, in particular its layer-by-layer versus three-dimensional growth [108], [109], [110], [111], [112]. Temperature also plays an important role in this calculation: relatively high temperatures are required
Concluding remarks
In this review, we have explored some of the issues associated with the surface chemistry of atomic layer deposition (ALD). We have summoned results and conclusions from our own work to illustrate some of the most important aspects of this chemistry. The main conclusion from the limited surface-science studies of these systems available to date is that the chemistry of the compounds used for ALD on the substrate is complex and deviates significantly from the idealized picture often envisioned
Acknowledgement
Financial support has been provided mainly by the U. S. Department of Energy, Materials Science Division, under Grant No. DE-FG02-03ER46599.
References (125)
Coord. Chem. Rev.
(1998)Prog. Mater. Sci.
(2003)Prog. Surf. Sci.
(2001)- et al.
Surf. Sci. Rep.
(2006) - et al.
J. Mol. Catal. A
(2008) Surf. Coat. Technol.
(2006)Surf. Sci.
(1991)- et al.
Surf. Sci.
(1994) - et al.
J. Cryst. Growth
(1991) - et al.
J. Cryst. Growth
(2011)
Appl. Surf. Sci.
Appl. Surf. Sci.
Polyhedron
Thin Solid Films
Thin Solid Films
Thin Solid Films
Thin Solid Films
Appl. Surf. Sci.
Thin Solid Films
Angew. Chem. Int. Ed.
Chem. Rev.
Chem. Vap. Deposition
Annu. Rev. Chem. Biomol. Eng.
J. Vac. Sci. Technol. B
J. Appl. Phys.
MRS Bull.
Corros. Rev.
Chem. Soc. Rev.
J. Phys. Chem. Lett.
Chem. Mater.
Rev. Sci. Instrum.
Appl. Phys. Lett
J. Phys. Chem. C
Chem. Rev.
J. Mater. Chem.
J. Kor. Phys. Soc.
Chem. Mater.
Chem. Mater.
J. Vac. Sci. Technol. A
J. Vac. Sci. Technol. A
J. Phys. Chem. B
J. Phys. Chem. A
ECS Trans.
J. Electrochem. Soc.
J. Phys. Chem. C
J. Phys. Chem. C
J. Vac. Sci. Technol. B
J. Vac. Sci. Technol. A
J. Phys. Chem. Lett.
J. Phys. Chem. Lett.
Cited by (91)
Time evolution of surface species during the ALD of high-k oxide on InAs
2023, Surfaces and InterfacesCoordination and organometallic precursors of group 10 and 11: Focused electron beam induced deposition of metals and insight gained from chemical vapour deposition, atomic layer deposition, and fundamental surface and gas phase studies
2022, Coordination Chemistry ReviewsCitation Excerpt :The loss of organic ligands as consequence of electron irradiation can be monitored by reflection absorption infrared spectroscopy (RAIRS) [103–105]. Mass spectrometry (MS) is also a powerful approach to study electron-stimulated desorption (ESD) of fragments that occurs upon irradiation of the condensed precursor monolayers or bulk surfaces as well as to assess thermally driven desorption of volatile products upon heating [106,107]. Note that most experiments monitor ESD of neutral species by post-ionization in the mass spectrometer.
Atomic layer deposition (ALD) assisting the visibility of metal-organic frameworks (MOFs) technologies
2021, Coordination Chemistry ReviewsTowards understanding the first half-ALD cycle of Ag growth: adsorption and dissociation of silver(i) acetamidinates on the Ag(110) surface
2024, Physical Chemistry Chemical PhysicsControlling electron beam induced deposition of iron from Fe(CO)<inf>5</inf>: Inhibition of autocatalytic growth by NH<inf>3</inf>and reactivation by electron irradiation
2023, Journal of Vacuum Science and Technology A: Vacuum, Surfaces and Films
- ☆
Invited contribution to Special Thematic Issue entitled “CVD and ALD Precursor Design and Application”.