JUL 2012: Atomic Layer Deposition: Self-terminating and Saturating gas-solid Chemical Reaction on the Substrate
ALD encorporates two: self-limiting and complementary reactions, sequentially and alternating leading to the depositing of a homogeneous and conformal layer growth. Conformal thin-film deposition is opted in the deposition of Seed Layer for Copper Interconnection, Barrier layer for preventing Cu-electromigration into the Low-K Dielectrics, Ultra-Pure Magnetic Material for Magnetic Random Access Memory (e.g. DRAM) Units (ITRS Roadmap till 2020), e.t.c. This article deals with detial investigation of ALD reaction conditions on the growth of Molelecule during the surface gas-solid surface reaction. ALD of Al2O3 by TMA and Waters is considered as an ideal ALD process, which would be investigated by exploring the surface activities occuring during each half-cyle of ALD process. Change is growth per cycle (gpc) with different experimental conditions (esp. Pressure inside the chamber, Chamber/Substrate Temperature, Precursor/Reactant Dosing Temperures) are major highlight of this article. At the end, the results from the different ALD Process conditions are summarized to extract very important information about the ALD Window necessary for getting a perfectly homogeneous and confrmal growth of sub-monolayer film.
Historically, ALD was introduced in year 1970. It used to be termed as Atomic Layer Epitaxy (ALE). Inital research on ALE was carried out by the Research Group from University of Helsinky, Finland led by Prof. Dr. Tuomo Suntola. In year 1977, Pro. Suntola filed a patent for the production of compound thin-films, which was the first patent in the development of ALD process . The first pioneering research work on ALE from Prof. Suntola was published in Materials Science Reports in 1989 . The inital phase of ALE was entirely dedicated towards the growth of single crystal of III-V and II-VI compounds and ordered heterostructures like superlattices and superalloys. Those days ALE was studied to meet the requirement for the improvement of ZnS thin-films and dielectric thin-films for electrolumniscent thin-film displays .
Fig. (1): Figure 3. Cross-sectional SEM image of an conforma Al2O3
ALD ﬁlm with a thickness of 300 nm on a Si-wafer with a trench structure. (Source: From ref 42. Copyright 1999 John Wiley & Sons.)
The actual reason behing conformal growth of sub-monolayer of gaseous molecule (precursor) on the substrate (solid-state) is due to self-terminating and saturating chemical reaction taking place on the substrate. The actual self-terminating behaviour has not yet been discovered in many ALD-Processes. Interestingly, huge research on ALD process of Alumina (Al2O3) on Silicon or Silicon Dioxide Substrate has already been carrier out leading to understanding behind the actual chmemical activitiy of Al2O3 ALD process. ALD of AL2O3 on Silicon substrate takes place by the adsorption of TMA on Silicon Substrate at the -OH terminated sites (Step 2, Fig. 2).
Fig 2. Self-terminating Chemiabsorption of TMA of the Substrate followed by the Ligand Exchange during Al2O3 ALD on Si-substrate.
This half-cycle produces -Me (Methyl terminated group) on the Al with the release of Methane gas. The steric hindrance by these -Me groups would block any further adsorption of TMA molecules on the surface. Hence, a single sub-monolayer of precursor molecule is absorbed on the substrate. After purging, methane gas along with the unreacted TMA are exhausted. In the next step (Step 3, fig. 2) after the introduction of Water, the ligand exchange between -Me group and -OH group takes place. Finally, the original substrate condition is again rejuvinated, termed as the complimentary reaction. Hence, a single layer of Aluminium Oxide is deposited on the substrate with the re-growth of original substrate state (-OH termination). This self-terminating behavior of ALD half-cycle reactions can be further supported by some schematic plots (fig. 6). This leads to the evolution of Random Growth of molecule as observed in ALD process. In random growth, the deposited material covers throughout the surface, which is the actual reason behind conformal material growth by the ALD process. However, the huge question rises about the experimental condition necessary to have such saturating half-cycle reactions for a ALD process. Hence, it is necessary to observe the actual material growth during each ALD cycle i.e growth per cycle (gpc) for different experimental conditions. For certain pressure and temperature conditions gpc remains constant which is the safe zone for ALD cycle termed as the ALD Window. Thus, for any ALD process it is outmost necessary to note the ALD temperature and pressure conditions.
Note: For the ALD process, it is necessary that the substrate temperature is below 220°C, so that inter-diffusion of material is prevented and the system can be maintained under controlled budget (next generation microelectronics application).
Fig. 3: Self-saturating half-cycle reaction depicted by a schematic diagram for the ALD of Al2O3 by TMA (Precursor) and Water (Reactant).
Saturative ALD reaction observed during the Experiment
The saturative ALD experiment can be best observed by plotting the amout the precursor molecule adsorbed on the surface (Step 2, Fig. 3) on the prepared sample (Step 1, Fig. 3). Hence, amount of Aluminium adosrved is plotted against -OH surface concentration in the following fig. 4. As expected, the amout of Aluminium linearly increases with increased concentration of surface -OH terminated sites.
Fig. 4: Plot of Aluminium adsorbed on the -OH terminated surface plotted for 1nm by 1nm area against -OH concentration .
Fig. 5: Plot of amount of Al atoms adsorbed (in 1nm by 1nm square)/nm2 area with increasing ALD deposition cycles .
The plot of amout of Al atoms adsorbed on differently treated Silicon/SiO2 substrate further supports the necessisity of -OH terminated substrate. The gpc increases in first few cycles in -OH terminated surface (Chemical SiO2) compared with the -H terminated Silcon substrate where it exhibits difficulty in initial nucleation (gpc is nearly zero). After 15 cycles in both type of substrates, the ALD reaction saturates with no further adsorption of Al-atoms on the substrate.
Exprimental ALD process of AL2O3 by TMA Precursor
In this section, the variation of reaction conditions on Al2O3 gpc would be separately studied. The ideal case for ALD cycle is depicted in fig. 4. It is worth noting that, within certain temperation range for the ALD Gpc is constant which is the ALD window. Within the ALD window the layer grwoth is linear with increasing ALD cycle which is the final optimized experimental conditions or parametres (Fig. 4(iv)).
Fig. 6: (i) Variation of gpc with substrate temperation (T): (a) the ALD Window, (b) condensation at low temperation, (c) incomplete reaction at low temperature, (d) decomposition at higher temperation and (e) desorption at higher temperature. (ii) Gpc chamge with purging time. (iii) Gpc change with precursor dosing time or precursor dosing amount: after certain amount of dose further precursor is not absorbed on the substrate, which is the onset of saturation of starting of purging cycle. (iv) The outcome of (1), (2) and (3) is necessary to get constant amout of precursor is absorved and we get constant gpc (slope of the curve) with increasing number of cycles.
(1) Gpc change with precursor dosing time i.e. precursor dosing amount. After certain amount of dose further precursor is not absorbed on the substrate. This behavior is also termed as the Irreversible Saturation Adsorption as observed in the ALD process.
(2) In addition, in rapid ALD process, the time lapse between the first precursor dosing time and the saturation time is drastically reduced.
Huge no. of experiments has already been performed with varying experimental condition for the ALD of Al2O3 from TMA precursor. Interestingly, the systemtic investigation of experimental conditions for the Al2O3 has been reported in the Scientific Article published by Prof. Dr. Riikka L. Puurunen from IMEC and VTT Technical Research Centre in year 2005. Detail surface chemistry of atomic layer deposition has been repored in this article titled- Surface chemistry of atomic layer deposition: A case study. With reference to this journal, variation of experimental condition on Al2O3 gpc is depicted in the following figures. It must be noted that the target is to have constant gpc during the ALD cycle.
(a) ALD Temperature Window
Fig. 7: Plot of amount of Al adsorbed with increasing ALD or substrate temperature .
Result: The ALD window is identified to be between 170 and 300 °C.
(b) ALD Pressure Window
Fig. 8: Calculated gpc (nm/cycle) is plotted against varying Vapor Pressure for constant Precursor and for constant Reactant. The experiment was carried out at 340°C .
Result: GPC is independent of varying Vapor Pressure of both precursor and reactant measured at constant precursor/reactant vapor pressure of 1E-4 Torr.
(b) Precursor/Reactant Dosing and Purging Window
Fig. 9: Growth per cycle or deposition per cycle (GPC) is plotted against (top) TMA pulsing and Purging time, (bottom) Water pulsing and purging time. The experiment was carried out at 300°C .
Result: The outcomes are only valid for the ALD of Al2O3 by TMA precursor. The following outcomes can be directly noted from the plot of gpc vs. time, shown in fig. 9. The minimum time for each half-cycle dosing/purging period is found out by the time scale when gpc value saturates. The minimum ALD precursor dosing time is 100 ms at 300°C. It is not recommended to increase the ALD precursor dosing time, because higher is the ALD precursor dosing time, in larger amount the precursor is wasted beyond the saturation. The minimum precursor purging time is observed to be 600 ms. Similarly, the minimum ALD reactant dosing time is observed to be 200 ms and the following minimum reactant purging time is 400 ms at 300°C. With different ALD temperature the observed value can vary.
(c) Precursor dosing temperature
Fig. 10: Plot between amout of Al adsorbed vs. Precursor dosing temperature .
Result: The amount of Al adsorbed on the surbstrate is fairly constant with temperature between 50 and 300 °C.
Note: If not mentioned, ALD temperature is same as the Substrate Temperature. However, the reactant and precursor temperature can be of different value.
The following outcome can be noted for the ALD of Al2O3 by TMA and Water/Hydrogen Peroxide Precursor. The results are depicted in the following table:
||Minimum time (ms)
|ALD Susbrate Conditions
|Precursor Dosing Conditions
||E-4 Torr at 300°C
||100 at 300°C
|Precursor Purging Conditions
||E-4 Torr at 300°C
||600 at 300°C
|Reactant Dosing Conditions
||E-4 Torr at 300°C
||400 at 300°C
|Reactant Purging Conditions
||E-4 Torr at 300°C
||200 at 300°C
* Depends on the Number of Cycle (film-thickness). Time is dependent on the Cycle Period.
The general Practice for ALD of Al2O3 by TMA and Water is done by the following values: -Precursor Dose (3s) - Purge (5s) - Reactant Dose (3s) - Purge (5s). ALD temp. is below 225 °C.
 T. Suntola, J. Antson, U.S. Patent, 4,058, 430, 1977.
 T. Suntola, Materials Science Reports, 4, 5, 261–312, 1989 .
 Suntola, T. Thirty Years of ALD. An Invited Talk at AVS Topical Conference on Atomic Layer Deposition (ALD2004); University of Helsinki: Helsinki, Finland, 2004.
 Ahonen, M.; Pessa, M.; Suntola, T. Thin Solid Films, 65, 301, 1980.
 Riikka L. Puurunen, J. Appl. Phys. 97 , 121301, 2005