Nonequilibrium molecular dynamics simulations of organic friction modifiers

NEMD Simulations

The requirement for greater energy efficiency in engineering systems has led to a general reduction in lubricant viscosity, which means that an increasing number of engineering components operate under boundary lubrication conditions. As a result, lubricant additives that reduce friction and wear under boundary conditions are of increasing importance [1]. Moreover, with growing concern that engine exhaust after-treatment systems may be poisoned by elements found in some types of friction modifier additives, there has been a resurgence of interest in organic friction modifier (OFM) additives which are based solely on C, H, O, and N atoms [2].

OFMs are amphiphilic surfactant molecules that contain a nonpolar hydrocarbon tail group attached to a polar head group. They reduce friction through adsorption of the polar head group to metal or ceramic surfaces, with strong, cumulative van der Waals forces between neighbouring tail groups leading to the formation of incompressible monolayers that prevent contact between solid surfaces to reduce adhesion and friction [1].

Nonequilibrium molecular dynamics (NEMD) simulations have been used to examine the atomistic structure and friction properties of commercially relevant organic friction modifier (OFM) monolayers adsorbed on iron oxide surfaces and lubricated by a thin, separating layer of hexadecane [2]. Specifically, carboxylic acid, amide, and glyceride OFMs, with saturated and Z-unsaturated hydrocarbon tail groups, are simulated at various surface coverages and sliding velocities. At low and medium coverage, the OFMs form liquid-like and amorphous monolayers, respectively, which are significantly interdigitated with the hexadecane lubricant, resulting in relatively high friction coefficients. At high coverage, solid-like monolayers are formed which, during sliding, results in slip planes between well-defined OFM and hexadecane layers, yielding a marked reduction in the friction coefficient. OFMs with glyceride head groups yield significantly lower friction coefficients than amide and particularly carboxylic acid head groups. For all of the OFMs and coverages simulated, the friction coefficient is found to increase linearly with the logarithm of sliding velocity; however, the gradient of this increase depends on the coverage. The structure and friction details obtained from these simulations agree well with experimental results and also shed light on the relative tribological performance of these OFMs through nanoscale structural variations. This has important implications in terms of the applicability of NEMD to aid the development of new formulations to control friction [2].

NEMD simulations

Fig 1. NEMD simulations of iron oxide surfaces lubricated by a thin layer of n-hexadecane and high (left) and medium (right) coverage OFM films.


A further NEMD study compared the behaviour between confined OFM systems modelled with cheaper united-atom force-fields (where nonpolar hydrogen atoms are grouped with carbon atoms) and all-atom force-fields (where hydrogen atoms are modelled explicitly) [3]. The study showed that, although united-atom force-fields are significantly cheaper, accurate, all-atom force-fields are required in order to obtain friction behaviour which agrees with experimental observations for these systems [3].

A recent NEMD study also investigated the impact of nanoscale surface roughness on the structure and friction of OFM films [4]. Different levels of 3-D nanoscale root-mean-squared (RMS) surface roughness were added to iron surfaces using a random midpoint displacement (RMD) algorithm. The direct contact of asperities was prevented under all the conditions simulated due to strong adsorption, which prevented squeeze-out. As for atomically-smooth surfaces [2], an increased coverage resulted in significantly lower friction forces to due more ordered, solid-like films with less interdigitation. Rougher surfaces led to more liquid-like, disordered films and higher friction. However, the effect of coverage is far stronger than the effect of roughness in the ranges studied. This study confirms that stearic acid films are almost as effective on contact surfaces with nanoscale roughness as those which are atomically-smooth [4].

NEMD Simulations

Fig. 2. NEMD simulations of iron surfaces with random, 3-D nanoscale roughness lubricated by high coverage OFM films.

Material was kindly provided by James Ewen, Tribology Group, Imperial College London.


  1. Spikes, H. Friction Modifier Additives. Tribol. Lett. 2015, 60, 5.
  2. Ewen, J. P.; Gattinoni, C.; Morgan, N.; Spikes, H.; Dini, D. Nonequilibrium Molecular Dynamics Simulations of Organic Friction Modifiers Adsorbed on Iron Oxide Surfaces. Langmuir 2016, 32, 4450–4463.
  3. Ewen, J. P.; Gattinoni, C.; Thakkar, F. M.; Morgan, N.; Spikes, H.; Dini, D. A Comparison of Classical Force-Fields for Molecular Dynamics Simulations of Lubricants. Materials. 2016, 9, 651.
  4. Ewen, J. P.; Echeverri Restrepo, S.; Morgan, N.; Dini, D. Nonequilibrium molecular dynamics simulations of stearic acid adsorbed on iron surfaces with nanoscale roughness. Tribol. Int. 2017, 107, 264–273.

Administration of the project


  1. “As for atomically-smooth surfaces, an increased coverage resulted in significantly lower friction forces to due more ordered, solid-like films with less interdigitation.”

    Nice to see that Hugh Spikes’ group factored surface roughness into the MD simulations, yet…

    “However, the effect of coverage is far stronger than the effect of roughness in the ranges studied. This study confirms that stearic acid films are almost as effective on contact surfaces with nanoscale roughness as those which are atomically-smooth.”

    Wanna’ bet on that in real life scenarios?

    • Hi Rick,

      Thanks for your interest in this work.

      Contact surfaces generally have roughness on several length scales, including the nanoscale. On the basis of this study, your highlighted statement can only be substantiated for OFM-lubricated systems with the levels of roughness accessible to MD simulations (nanoscale RMS). Unlike nonpolar lubricant molecules, OFM molecules strongly adsorb to contact surfaces and high coverage films can ‘blur’ nanoscale surface roughness to yield smooth sliding interfaces and thus low friction.

      As far as I’m aware, the effect of other roughness scales (e.g. microscale) on the friction in OFM-lubricated systems has not been studied in boundary friction experiments. However, since the roughness features considered in these simulations are on the atomic scale, they have the most potential to weaken OFM films by disrupting intermolecular forces between neighbouring molecules. One may tentatively conclude, therefore, that they are the most important roughness scale in this particular case.

  2. Hello James,

    I have two fundamental problems with these types of tribological studies.

    While recognizing the practical computational limits of most MD simulations (unless, of course, you have access to MIRA or the like), those limitations (nanoscale roughness) have consequences on the value of the conclusions derived therefrom. After all, these studies are aimed at aiding in the development of new tribological approaches to real world friction problems. MD simulations where the asperity depths are a mere fraction of the adhered polar molecule chain length are not going to provide meaningful insight into how these films will perform in the field. The limitations of the computing power should not dictate the design of the MD study. Design an accurate MD model and apply for access to a bigger machine to crunch your numbers.

    Further, although I wholeheartedly agree that OFMs are very important in tribology, I suggest that (real world) asperity interactions produce temperature conditions that will completely destroy any surface-adhered OFM. There is little cause for concern about disruption of the intermolecular forces between the neighboring chains when the films themselves are vaporized by the heat of the friction event.

    This brings me to the recurring point regarding the overwhelming importance of ISN surface perfection technology.

    Regardless of the tribocoating methods employed, the asperities always win in the end! The only lasting solution to mechanical friction is the permanent removal of the surface asperities, which can then open the door for utilization of various clever friction-reducing tribofilms.

    Considering the aforegoing, is the provided nanoscale roughness of the OFM film in the MD simulation really the “most important” in this case? Does this not skew all the conclusions derived therefrom?

    • Whilst I fully appreciate your concerns in this respect, even simulations of this size require fairly extensive hardware (many 100s-1000s CPUs). Indeed, the supercomputer I run my MD simulations on is currently around 394th in the TOP500 list. Simulating surfaces with greater levels of roughness requires larger simulation cells in order to provide a realistic ‘steepness’ of the roughness features, making them even more expensive. Although the parallelisation of the code I use if fairly good, above a certain point, throwing more CPUs at the problem doesn’t always give a big performance benefit. Thus, at present, this is around the practical limit for all-atom NEMD simulations of realistic rough surfaces. I would strongly dispute that MD simulations at this scale can provide little meaningful insight to the behaviour of these systems.

      On the contrary, extensive research by Prof. Spikes and others on realistic contact surfaces show that OFM films significantly reduce boundary friction under even the most extreme conditions (GPa pressures). This suggests that OFM films form a strong monolayer which effectively minimises direct solid-solid contact and thus prevents the extreme temperature rises to which you refer.

      Though such technologies certainly show significant promise in significantly reducing friction and wear, they are some distance from being proven as a universal lubrication solution. OEMs are understandably extremely wary of any additive which effectively promotes wear in order to polish surface asperities. Are these additives removed after an initial wear-in period? Otherwise, what is to stop these additives promoting wear after this period where they clearly provide a benefit. These kinds of issues are discussed in more detail in Prof. Spikes’ Friction Modifier review (reference 1 above).

      For now, your last point remains conjecture, but its certainly something we are interested in investigating in the future.

  3. Well James, 394th ain’t so bad; it’s still way more computing power than most have access to, including me. 😉

    As you are already aware, recent studies have shown that the asperity “steepness” angle is the most significant factor in determining the resulting friction therefrom. So, as said earlier, larger MD simulations (larger simulation cells) would naturally yield more realistic data. With this point, I’m sure you agree – expenses be damned.

    Regarding the limited value (insight) comment I made, that comes in regard to the inconvenient truth that over enough time, the cumulative tribofilm failures (on the nanoscale) will inevitably result in macro-scale damage to the system. No tribofilm is going to survive (on the nanoscale) the heat generated during events capable of melting the asperity tips into molten metal.

    In the latest generation of ISN technology (ISS – in-situ synthesis), the nanopolishing agents are synthesized within the system as it operates from chemical precursors in the lubricant. These agents remain on the surfaces (exhibiting ferromagnetism) and do not continue to abrade the surface once atomic-level perfection is achieved.

    As for the proven efficacy of ISN technology, it’s not so far away as you might think…

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.