Atomic layer deposition

from Wikipedia, the free encyclopedia

The atomic layer deposition ( English atomic layer deposition , ALD ) is a method for the deposition of extremely thin layers , down to atomic monolayers , on a starting material.

It is a heavily modified chemical vapor deposition ( CVD ) process with two or more cyclical self-limiting surface reactions . The material to be deposited is bound in chemical form to one or more carrier gases, the so-called precursors . These precursors are fed alternately into a reaction chamber and made to react there with the substrate, whereupon the substance bound in the gas is deposited on the substrate material. The resulting layers usually have a polycrystalline or amorphous structure. For single crystal ( epitaxial ) layer, the process is also known as atomic layer epitaxy (engl. , Atomic layer epitaxy , ALE ) is known.

If it is not individual atoms from the predecessor molecules but rather molecular fragments that are deposited in a self-limiting reaction, one also speaks of molecular layer deposition (MLD, " molecular layer deposition ").

Different names, one principle, atomic layer deposition
designation abbreviation
Atomic layer deposition ALD
Atomic layer epitaxy ALE
Atomic layer evaporation ALE
Atomic layer growth ALG
Chemical assembly
Molecular deposition
Molecular lamination
Molecular layer deposition MLD
Molecular layer epitaxy MLE
Molecular layering ML
Molecular stratification

history

ALD was developed twice independently: Once (Engl., Under the name "molecular Agen stratification" molecular layering , ML ) in the 1960s in the Soviet Union and the mid-1970s under the name "atomic layer epitaxy" (English. Atomic layer epitaxy , ALE ) in Finland . Due to the global block formation after the Second World War and the long-term neglect of Soviet research results after the collapse of the Eastern Bloc , the work on the layering of molecules was forgotten for years. In order to elucidate the early history, a virtual project on the history of the ALD process (VPHA - virtual project on the history of ALD ) was formed, which was launched in summer 2013 by a group of scientists. The results of the VPHA are essays that describe the historical development of ALE and ML, a review of the early publications on the topic that are most worth reading up to 1986, and an article on the findings of the VPHA.

At the time, was looking for a method to produce high-quality coatings on large area substrates, such as thin-film electroluminescence Indicators (Engl. Thin-film electroluminescent , TFEL ).

In the 1980s, the prospect of using ALD for epitaxial semiconductor layers also led to large investments in this area. Due to the chemical incompatibility of alkyl compounds of main group III and hydrides of main group V, ALD did not bring any real advantages over molecular beam epitaxy (MBE) or organometallic gas phase epitaxy (MOVPE).

It was not until the mid-1990s that ALD received more attention as a promising coating technology in microelectronics . The main reasons for this are the progressive reduction in structure and higher demands on the aspect ratios in integrated circuits and the associated search for new materials and deposition techniques. Only a few applications aim at the deposition of epitaxial layers. The very thin layers (approx. 10 nm) are often of an amorphous structure.

principle

Schematic representation of an ALD reaction cycle for an ALD system made up of two components.

As with other CVD processes, ALD also creates the layer through a chemical reaction of at least two starting materials ( precursors ). In contrast to conventional CVD processes, with ALD the starting materials are let into the reaction chamber one after the other. The reaction chamber is normally flushed with an inert gas (e.g. argon ) between the gas inlets for the starting materials . In this way, the partial reactions should be clearly separated from one another and limited to the surface. An essential feature of ALD is the self-limiting character of the partial reactions, i.e. the starting material of a partial reaction does not react with itself or with ligands of itself, which limits the layer growth of a partial reaction to a maximum of one monolayer for any length of time and amount of gas .

The simplest ALD procedure is a two-component system, e.g. B. for tantalum (V) oxide (Ta 2 O 5 ) the components tantalum pentachloride (TaCl 5 ) and water (H 2 O). As described above, the two components are now fed into the chamber alternately and separately through rinsing steps. The following four characteristic steps result:

  1. A self-limiting reaction of the first reactant (Reactant A, TaCl 5 )
  2. A purge or evacuation step (of the reaction chamber) to remove unreacted gas of the first reactant and reaction products
  3. A self-limiting reaction of the second reactant (Reactant B, H 2 O) or another step (e.g. plasma treatment) to reactivate the surface for the first reaction
  4. A flush or evacuation step

These four steps are summarized in a so-called (reaction) cycle, which has to be repeated several times during the coating process in order to achieve the desired layer thickness. Ideally, each action step is complete; That is, the precursor molecules chemisorb or react with the surface groups until the surface is completely covered. Thereafter, no further adsorption takes place (self-limitation). The layer growth is self-controlling or self-limiting under these reaction conditions; That is, the amount of layer material deposited in each reaction cycle is constant.

Depending on the process and reactor, a cycle lasts between 0.5 and a few seconds, with 0.1 to 3 Å of film material being generated per cycle (strongly dependent on the material system and the process parameters). In reality, this means that a closed layer of the target material cannot be reached in one cycle, so the term atomic layer deposition can be somewhat misleading. There are two main reasons for the reduced separation rate - usually given in GPC, which stands for growth per cycle (Eng. 'Growth per cycle'):

  1. The spatial expansion of the starting materials or their adsorbed ligands results in what is known as steric hindrance , which has the effect that parts of the surface are shielded and thus prevent adsorption at this point.
  2. Incomplete partial reactions, as a result of which fewer reaction points are available on the surface, i.e. areas that cannot participate in the partial reaction.

In addition, etchbacks can also occur, for example through the reaction products of halogen-containing precursors or through ion bombardment when using (direct) plasma.

Advantages and disadvantages

Despite the non-ideal growth in real processes, there are several advantages in summary when depositing thin layers using ALD. An essential point is the very good layer thickness control of ultra-thin layers smaller than 10 nm. Because of the self-limiting reaction, the layer only grows by a determinable value per cycle, which is independent of the cycle duration in the saturation range. The layer grows proportionally to the number of reaction cycles, which enables precise control of the layer thickness. An exception is the beginning of the coating, at which a faster or slower growth can take place due to a possibly different surface chemistry of the substrate material.

The separate dosing of the precursor substances prevents undesired gas phase reactions in the sample space and also enables the use of highly reactive precursors. Due to the fixed dosage, each reaction step has enough time to complete, which means that high-purity layers can also be produced at relatively low temperatures. In addition, compared to other CVD processes, the requirements for the homogeneity of the gas flow are significantly lower. Therefore, the ALD is theoretically particularly suitable for coating large areas or more structured surfaces. In practice, however, economic aspects, such as high throughput and low gas consumption, play a decisive role. For these reasons, a uniform gas distribution is nevertheless necessary.

The disadvantage is the very slow speed of the process. Aluminum oxide can be grown with 0.11 nm per step, for example. Depending on the requirements for the resulting material quality, 100–300 nm layers per hour can be produced using ALD. In addition, the demands on the quality and purity of the substrate are very high if you value the most monocrystalline ALD layers possible. ALD devices are themselves very expensive and the manufacturing costs of a shift vary widely. For many materials it is difficult to find suitable precursors (starting materials) that are reactive enough for the ALD process, but do not thermally decompose too quickly. The precursors are therefore often expensive, have to be stored refrigerated and are often only usable for a few months.

literature

  • Esko Ahvenniemi, Andrew R. Akbashev, Saima Ali, Mikhael Bechelany, Maria Berdova, Stefan Boyadjiev, David C. Cameron, Rong Chen, Mikhail Chubarov: Review Article: Recommended reading list of early publications on atomic layer deposition — Outcome of the “Virtual Project on the History of ALD " . In: Journal of Vacuum Science & Technology A . tape 35 , 2017, p. 010801 , doi : 10.1116 / 1.4971389 .
  • Riikka L. Puurunen: Surface chemistry of atomic layer deposition: A case study for the trimethylaluminum / water process . In: Journal of Applied Physics . tape 97 , no. 12 , 2005, p. 121301-01 , doi : 10.1063 / 1.1940727 .

Web links

Individual evidence

  1. ^ Riikka L. Puurunen: Surface chemistry of atomic layer deposition: A case study for the trimethylaluminum / water process . In: Journal of Applied Physics . tape 97 , no. 12 , 2005, p. 121301-01 , doi : 10.1063 / 1.1940727 .
  2. ↑ Patent US4058430 : Method for producing compound thin films. Applied on November 25, 1975 , published November 15, 1977 , inventors: T. Suntola , J. Antson. ‌
  3. ^ Riikka L. Puurunen: A Short History of Atomic Layer Deposition: Tuomo Suntola's Atomic Layer Epitaxy . In: Chemical Vapor Deposition . tape 20 , 2014, p. 332-344 , doi : 10.1002 / cvde.201402012 .
  4. ^ Anatolii A. Malygin, Victor E. Drozd, Anatolii A. Malkov, Vladimir M. Smirnov: From VB Aleskovskii's "Framework" Hypothesis to the Method of Molecular Layering / Atomic Layer Deposition " . In: Chemical Vapor Deposition . Volume 21 , no. 10-11-12 , 2015, pp. 216–240 , doi : 10.1002 / cvde.201502013 .
  5. ↑ Esko Ahvenniemi, Andrew R. Akbashev, Saima Ali, Mikhael Bechelany, Maria Berdova, Stefan Boyadjiev, David C. Cameron, Rong Chen, Mikhail Chubarov: Review Article: Recommended reading list of early publications on atomic layer deposition — Outcome of the “ Virtual Project on the History of ALD " . In: Journal of Vacuum Science & Technology A . tape 35 , 2017, p. 010801 , doi : 10.1116 / 1.4971389 .
  6. ^ Riikka L. Puurunen: Learnings from an Open Science Effort: Virtual Project on the History of ALD . In: ECSarXiv. 2018, doi : 10.1149 / 08606.0003ecst .