Vol. 25, No.

9 | 1 May 2017 | OPTICS EXPRESS 9778

Interference effects in laser-induced plasma

emission from surface-bound metal micro-
particles
EYAL FEIGENBAUM,1,* OMER MALIK, 1 ALEXANDER M. RUBENCHIK, 1 AND
MANYALIBO J. MATTHEWS1
1
National Ignition Facility and Photon Science, Lawrence Livermore National Laboratory, Livermore,
CA, 94550, USA
*eyal@llnl.gov

Abstract: The light-matter interaction of an optical beam and metal micro-particulates at the
vicinity of an optical substrate surface is critical to the many fields of applied optics.
Examples of impacted fields are laser-induced damage in high power laser systems, sub-
wavelength laser machining of transmissive materials, and laser-target interaction in directed
energy applications. We present a full-wave-based model that predicts the laser-induced
plasma pressure exerted on a substrate surface as a result of light absorption in surface-bound
micron-scale metal particles. The model predictions agree with experimental observation of
laser-induced shallow pits, formed by plasma emission and etching from surface-bound metal
micro-particulates. It provides an explanation for the prototypical side lobes observed along
the pit profile, as well as for the dependence of the pit shape on the incident laser and particle
parameters. Furthermore, the model highlights the significance of the interference of the
incident light in the open cavity geometry formed between the micro-particle and the
substrate in the resulting pit shape.
© 2017 Optical Society of America
OCIS codes: (140.3330) Laser damage; (240.0240) Optics at surfaces; (160.4670) Optical materials; (290.0290)
Scattering; (140.3390) Laser materials processing.

References and links

1. S. Demos, M. Staggs, K. Minoshima, and J. Fujimoto, “Characterization of laser induced damage sites in optical
components,” Opt. Express 10(25), 1444–1450 (2002).
2. B. Bertussi, P. Cormont, S. Palmier, P. Legros, and J. L. Rullier, “Initiation of laser-induced damage sites in
fused silica optical components,” Opt. Express 17(14), 11469–11479 (2009).
3. T. I. Suratwala, P. E. Miller, J. D. Bude, W. A. Steele, N. Shen, M. V. Monticelli, M. D. Feit, T. A. Laurence, M.
A. Norton, C. W. Carr, and L. L. Wong, “HF-based etching processes for improving laser damage resistance of
fused Silica optical surfaces,” J. Am. Ceram. Soc. 94(2), 416–428 (2011).
4. J. Bude, P. Miller, S. Baxamusa, N. Shen, T. Laurence, W. Steele, T. Suratwala, L. Wong, W. Carr, D. Cross,
and M. Monticelli, “High fluence laser damage precursors and their mitigation in fused silica,” Opt. Express
22(5), 5839–5851 (2014).
5. I. L. Bass, G. M. Guss, M. J. Nostrand, and P. L. Wegner, “An improved method of mitigating laser-induced
surface damage growth in fused Silica using a rastered, pulsed CO2 laser,” Proc. SPIE 7842, 784220 (2010).
6. M. J. Matthews, S. T. Yang, N. Shen, S. Elhadj, R. N. Raman, G. Guss, I. L. Bass, M. C. Nostrand, and P. J.
Wegner, “Micro-shaping, polishing, and damage repair of fused silica surfaces using focused infrared laser
beams,” Adv. Eng. Mater. 17(3), 247–252 (2015).
7. L. Lamaignère, M. Chambonneau, R. Diaz, R. Courchinoux, and T. Donval, “Laser damage resistance
qualification of large optics for high power Laser,” Proc. SPIE 9345, 934508 (2015).
8. C. W. Carr, D. A. Cross, M. A. Norton, and R. A. Negres, “The effect of laser pulse shape and duration on the
size at which damage sites initiate and the implications to subsequent repair,” Opt. Express 19(S4 Suppl 4),
A859–A864 (2011).
9. J. Wong, J. L. Ferriera, E. F. Lindsey, D. L. Haupt, I. D. Hutcheon, and J. H. Kinney, “Morphology and
microstructure in fused silica induced by high fluence ultraviolet 3ω (355 nm) laser pulses,” J. Non-Cryst. Solids
352(3), 255–272 (2006).
10. R. A. Negres, G. M. Abdulla, D. A. Cross, Z. M. Liao, and C. W. Carr, “Probability of growth of small damage
sites on the exit surface of fused silica optics,” Opt. Express 20(12), 13030–13039 (2012).
11. T. A. Laurence, J. D. Bude, N. Shen, T. Feldman, P. E. Miller, W. A. Steele, and T. Suratwala, “Metallic-like
photoluminescence and absorption in fused silica surface flaws,” Appl. Phys. Lett. 94(15), 151114 (2009).

#286889 https://doi.org/10.1364/OE.25.009778
Journal © 2017 Received 16 Feb 2017; accepted 12 Apr 2017; published 19 Apr 2017
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9779

12. P. E. Miller, J. D. Bude, T. I. Suratwala, N. Shen, T. A. Laurence, W. A. Steele, J. Menapace, M. D. Feit, and L.
L. Wong, “Fracture-induced subbandgap absorption as a precursor to optical damage on fused silica surfaces,”
Opt. Lett. 35(16), 2702–2704 (2010).
13. C. W. Carr, J. D. Bude, and P. DeMange, “Laser-supported solid-state absorption fronts in silica,” Phys. Rev. B
82(18), 184304 (2010).
14. E. Feigenbaum, S. Elhadj, and M. J. Matthews, “Light scattering from laser induced pit ensembles on high
power laser optics,” Opt. Express 23(8), 10589–10597 (2015).
15. E. Feigenbaum, N. Nielsen, and M. J. Matthews, “Measurement of optical scattered power from laser-induced
shallow pits on silica,” Appl. Opt. 54(28), 8554–8560 (2015).
16. S. Palmier, I. Tovena, L. Lamaignère, J. L. Rullier, J. Capoulade, B. Bertussi, J. Y. Natoli, and L. Servant,
“Study of laser interaction with aluminum contaminant on fused silica,” Proc. SPIE 5991, 59910R (2005).
17. B. S. Luk’yanchuck, Laser Cleaning (World Scientific, 2002)
18. M. J. Matthews, N. Shen, J. Honig, J. D. Bude, and A. M. Rubenchik, “Phase modulation and morphological
evolution associated with surface-bound particle ablation,” J. Opt. Soc. Am. B 30(12), 3233–3242 (2013).
19. R. N. Raman, S. G. Demos, N. Shen, E. Feigenbaum, R. A. Negres, S. Elhadj, A. M. Rubenchik, and M. J.
Matthews, “Damage on fused silica optics caused by laser ablation of surface-bound microparticles,” Opt.
Express 24(3), 2634–2647 (2016).
20. C. D. Harris, N. Shen, A. M. Rubenchik, S. G. Demos, and M. J. Matthews, “Characterization of laser-induced
plasmas associated with energetic laser cleaning of metal particles on fused silica surfaces,” Opt. Lett. 40(22),
5212–5215 (2015).
21. S. G. Demos, R. A. Negres, R. N. Raman, N. Shen, A. M. Rubenchik, and M. J. Matthews, “Mechanisms
governing the interaction of metallic particles with nanosecond laser pulses,” Opt. Express 24(7), 7792–7815
(2016).
22. P. M. Celliers, K. G. Estabrook, R. J. Wallace, J. E. Murray, L. B. Da Silva, B. J. Macgowan, B. M. Van
Wonterghem, and K. R. Manes, “Spatial filter pinhole for high-energy pulsed lasers,” Appl. Opt. 37(12), 2371–
2378 (1998).
23. E. Feigenbaum, R. N. Raman, D. Cross, C. W. Carr, and M. J. Matthews, “Laser-induced Hertzian fractures in
silica initiated by metal micro-particles on the exit surface,” Opt. Express 24(10), 10527–10536 (2016).
24. A. C. Fischer-Crips, Introduction to Contact Mechanics, 2nd ed. (Springer, 2007)
25. K. L. Johnson, Contact Mechanics (Cambridge, 1985)
26. D. Grojo, L. Boarino, N. De Leo, R. Rocci, G. Panzarasa, P. Delaporte, M. Laus, and K. Sparnacci, “Size scaling
of mesoporous silica membranes produced by nanosphere mediated laser ablation,” Nanotechnology 23(48),
485305 (2012).
27. C. R. Phipps, Jr., T. P. Turner, R. F. Harrison, G. W. York, W. Z. Osborne, G. K. Anderson, X. F. Corlis, L. C.
Haynes, H. S. Steele, K. C. Spicochi, and T. R. King, “Impulse coupling to targets in vacuum by KrF, HF, and
CO2 single-pulse lasers,” J. Appl. Phys. 64(3), 1083–1096 (1988).
28. I. R. Çapoglu and G. S. Smith, “A Total-Field/Scattered-Field Plane-Wave Source for the FDTD Analysis of
Layered Media,” IEEE Trans. Antenn. Propag. 56(1), 158–169 (2008).
29. E. D. Palik, Handbook of Optical Constants of Solids (Academic Press, 1985).
30. R. Fabbro, J. Fournier, P. Ballard, D. Devaux, and J. Virmont, “Physical study of laserproduced plasma in
confined geometry,” J. Appl. Phys. 68(2), 775–784 (1990).
31. L. Berthe, R. Fabbro, P. Peyre, L. Tollier, and E. Bartnicki, “Shock waves from a water-confined laser-generated
plasma,” J. Appl. Phys. 82(6), 2826–2832 (1997).
32. W. Dieter, Bäuerle, Laser Processing and Chemistry, 4th ed. (Springer, 2011)
33. J. E. Sinko and C. R. Phipps, “Modeling CO2 laser ablation impulse of polymers in vapor and plasma regimes,”
Appl. Phys. Lett. 95(13), 131105 (2009).
34. W. H. Krueger and S. R. Pollack, “The initial oxidation of aluminum thin films at room temperature,” Surf. Sci.
30(2), 263–279 (1972).
35. O. Grushko, B. Ovsyannikov, and V. Ovchinnikov, Aluminum-Lithium Alloys: Process metallurgy, Physical
metallurgy and Welding (CRC Press, 2016).
36. G. L. Klimchitskaya, U. Mohideen, and V. M. Mostepanenko, “Casimir and van der Waals forces between two
plates or a sphere (lens) above a plate made of real metals,” Phys. Rev. A 61(6), 062107 (2000).
37. L. Bergström, “Hamaker constants of inorganic materials,” Adv. Colloid Interface Sci. 70, 125–169 (1997).

1. Introduction
The need for optics that can sustain higher laser fluences is driven by the recent
advancements in energetic laser systems and their applications. A substantial effort has been
spent in the last decade on mitigating damage initiation in high power lasers optics and on the
suppression of the shot-to-shot growth of damage sites [1–13]. These ‘classic damage sites’
are initiated by sub-bandgap absorption in fractures located in a few hundred nanometer thick
sub-surface layer [10–12], and can therefore be effectively treated by optimized HF etching
[3,4]. Recently, more effort has been focused on studying laser damage initiated by
particulate contamination [14–23], in contrast to the as-fabricated fracture initiation of the
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9780

classic damage sites. One type of artifact initiated by the presence of metal particle
contamination on the exit surface of silica is sub-micron deep laser-induced shallow pits
(LSPs) [14–16]. Metal particulates can originate from ablated optical beam dumps or abrasion
between metal surfaces (e.g. aluminum) in the laser system. These LSPs are free of fracture
and do not change morphology under subsequent laser pulses (i.e. no growth). Nevertheless,
they appear on optics in high densities and lead to optics lifetime limitation through gradual
increase in scattering and reduced transmitted energy on target. The depth of a LSP is
typically less than 1 μm with a depth profile that is Gaussian-like for the narrower ones, but
with a formation of a puzzling ‘shoulder’ within the pit for the wider ones (see Fig. 1(a))
[14,20,21]. Previous work considered and ruled out the formation of this pit profile as a result
of densification [19]. A typical LSP morphology is also found to be axisymmetric with a
correlation between the depth and the width [14,21], and with the depth increasing with
incident fluence [19,21]. The far-field intensity distribution of the scattered light is
determined by the morphology of the LSP on the optics [14], which motivates further study of
the shape-governing formation mechanism (e.g., the scattering could result in pinhole closure
in high power lasers spatial filter [22]).
Unlike damage precursors embedded within the glass surface, in the case of LSP creation,
the incident laser beam deposits heat into the skin depth of the metal particle, which in turn
ejects a plasma plume towards the substrate surface, propelling the particle away from the
substrate, and etching a LSP into it [20,21] (see illustration in Fig. 1(b)). A similar
mechanism, but at higher incident fluence can result in laser-induced Hertzian fractures, due
to the expanding plasma exerting pressure on the silica surface [23], taking the role of a solid
blunt indenter in traditional Hertzian fractures [24,25]. These laser-induced fractures have the
propensity to grow with subsequent laser shots.

Fig. 1. A prototypical depth profiles of a narrow (red) and a wide (blue) light-induced shallow
pit (LSP) – the wider one showing a ‘shoulder’ formation deviation from a Gaussian-like
shape, with a 3D morphology rendering of a wide pit as an inset (based on results from [14]).
(b) Rendering of the laser-induced plasma emission from a metal particle on the exit surface of
a glass window: the heat absorbed in the metal sphere due to light incident from below creates
plasma which expands towards the surface (in red) and propels the particle away from the
surface (in black). (c) The configuration of the numerical simulation geometry. The calculated
EM fields are recorded within the red bounding rectangle.

Modeling the plasma emission process from a metal micro-particle bound to the exit
surface of a silica glass window is of a coupled multi-physics nature. The interference of the
incident light in the wavelength-scale open cavity formed between the metal micro-sphere
and the substrate determines the intensity distribution at the skin-depth layer of the metal
particle. This intensity distribution results in a corresponding energy absorption and
subsequent heat flow, based on the metal optical and thermal properties, which in turn result
in plasma emission. Here, we propose a model that uses a rigorous approach for calculating
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9781

the light intensity, being the source term of this multi-physics problem, along with a more
simplified approach for the heat dissipation and the plasma emission. For example, this model
does not include heat diffusion or the effect of plasma emission on the incident light, and the
justifications for these assumptions and their implications will be discussed. Nevertheless,
this approach emphasizes the light interference contributions to the resulting net effect, and
the good agreement with experimental results highlights the centrality of the interference in
the formation of the LSPs profile, especially with respect to the ‘shoulder’ pattern that
appears within the pit. This semi-analytical cascaded approach gives a clear view of how
geometrical variations affect the plasma emission. A better understanding of the plasma
emission from metal particles in the vicinity of a glass surface could be applied beyond the
field of laser damage to potential applications in sub-wavelength laser machining of
transmissive materials through controlled plasma jet emission [26]. The plasma emission
from metal micro and nano structures on a substrate in the context of laser-target scaling is of
importance to the fields of directed energy and target design for inertial confinement fusion
[27].
The methodology used to study nanoscale ablation pits will be described in section 2,
exemplified with step-by-step plots of the physical quantities in the serial model chain for one
arbitrarily chosen case. The important role that interference plays in determining the source
term of this system becomes evident already in this first example. In section 3 we will discuss
the validity of the model from two perspectives: the usage of a two-dimensional (2D)
numerical calculation, and the validity of the other multi-physics component simplifications
in the model for various circumstances. In section 4 we will present the results of a parametric
study of the system.
2. Model methodology
2.1 Calculating the EM fields
Finite-difference time-domain (FDTD) simulation of Maxwell’s equations (Lumerical
Solutions Inc.) of the configuration illustrated in Fig. 1(c) was conducted. Perfectly matched
layer (PML) boundary layers and a total-field/scattered-field [28] domain were used and
tested to reduce reflections from boundaries. The calculated fields were recorded at free-space
wavelength λ0 = 351 nm within the rectangle bounding the lower particle hemisphere (red
rectangle in Fig. 1(c)). Using Fourier analysis the continuous wave fields at a given
wavelength are calculated from the temporal short pulse calculation. The calculations are 2D
in the x-y plane (we will discuss the validity in section 3), studied separately with P-polarized
(E-field in the x-y plane) and S-polarized (E-field normal to the x-y plane) excitations. The
surface bound particle system is that of an Al sphere positioned on a surface of a silica glass
slab substrate, where the material permittivity is based on tabulated data [29]. The excitation
is from below (at incidence angle α = 0) unless otherwise mentioned.
2.2 Translating the calculated intensity to emitted plasma flux
The absorbed power is calculated based on the recorded E-field values using
Pabs = 2 ω Im ( ε ) E , where E is the electric field, ε is the complex permittivity, and ω is the
1 2

radial frequency. Since the glass absorption is negligible, the power is absorbed only in the
metal particle. The FDTD calculation is linearly dependent in the excitation field amplitude,
which allows renormalizing the results for excitation fluence at-will by multiplying the
resulting absorbed power distribution by the ratio between the desired excitation and the
recorded excitation intensities. We translate the absorbed power (Pabs) to absorbed energy
density (Eabs) by multiplying the former with an effective pulse length of 1 ns (constant
throughout this work). The conversion to energy units is required for comparison to the
required enthalpy for ablation at later steps of the model. Eabs is depicted in Fig. 2(a) for an Al
sphere of radius R = 1 μm on the exit surface of silica glass, for an incident light input fluence
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9782

of Φin = 1 J/cm2 – illustrating that the light is absorbed at the skin of the particle and
exponentially decays radially towards the sphere center.

Fig. 2. The simplified plasma emission model (results are for an Al sphere, radius R = 1 μm,
input fluence Φin = 1 J/cm2): (a) absorbed energy (log scale, P-polarized); (b) decay length
(black dash: predicted value from the Fresnel equations for a flat surface), (c) percentage of the
absorbed fluence (from the incident fluence), (d) ablated layer thickness, (e) ablated material
energy flux as a function of θ ; (f) emitted plasma pressure projected on the glass surface as a
function of x (on surface) normalized by the sphere radius R. The legend in (b) holds also for
(c)-(f). The curves are plotted for P and S polarizations, with and without the glass surface.
Note that the θ angular span of the lowest hemisphere is [-π,0].

The data points are then represented on a radial coordinate system and the power inside
small azimuthal angle bins are radially fit to exponentially decaying curves. The data near the
sphere edge suffer from stair-casing numerical errors of the Cartesian numerical
representation, and the data inside the sphere is negligible thereby sensitive to numerical
noise. Therefore, the fit is done using linear regression over a segment that lies between 2 to 4
decay lengths under the sphere surface. The reference decay length is calculated from the
Fresnel equation for a flat surface: Ldec = λ0 (4π Im(n)) , where n is the complex refractive index
of the metal sphere. The maximal calculation resolution is 2 nm (used in the metal segment
and immediate vicinity), with a non-uniform rectangular grid that reduces to 16 nm in air
segments away from the metal sphere. Numerical convergence studies were used to reach
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9783

these values. The azimuthal bins size used is π/200. The number of bins was chosen to be
high enough to represent the spatial variation on the sphere surface but still small enough to
smooth numerical artifacts (an azimuthal bin represents ~15 nm arc length for a 1 μm metal
sphere). For each azimuthal bin the linear regression results in the absorbed energy density on
the sphere surface (at r = R), Eabs R
, and the decay length, Ldec. The decay length is plotted in
Fig. 2(b) as a function of θ for both excitation polarizations and in two configurations: when
the particle is directly on an exit surface or is in space. The decay length is very similar for
the four cases, and deviates only a few angstroms from the ~6.5 nm value predicted from the
Fresnel equations for a flat surface (black dashed line), with a slightly slower decay in the
region under the particle. Integrating radially the absorbed energy density gives the absorbed
fluence: Φ abs = EabsR
⋅ Ldec .
The absorbed fluence as a percentage of the incidence fluence is plotted in Fig. 2(c) as a
function of θ. The results show that about 10% of the incident energy gets absorbed,
becoming in turn the source term for the plasma emission process. The azimuthal behavior of
the absorption for the two polarizations is distinctively different: for the S polarization the
absorption is mainly under the particle whereas for the P polarization there is enhanced
absorption off-to the side (at about 45°), a trend which agrees with the Fresnel equation
dependency on the incident angle on a flat interface. The interference effect of the reflection
from the metal particle and the glass surface interface is evident from the difference between
the different curves corresponding to cases with and without the glass surface: the glass
interface causes the absorption within the central region under the particle (about [-π/4, π/4])
to be more focused under a smaller region (about [-π/8, π/8]). This illustrates that the light
interference in the open cavity between the particle and the substrate surface results in
pronounced azimuthal side lobes in the absorbed energy profile on the metal particle. This
non-uniform absorption distribution due to interference is a major finding of this study.
The ablated material depth is calculated based on comparing the local energy density with
the ablation enthalpy threshold value. The ablated layer depth is calculated based on the radial
exponential decay into the sphere of the absorbed energy density Labl = Ldec ⋅ log( Eabs R
/ H abl ) and is

depicted in Fig. 2(d), where the ablation enthalpy for Al is Habl = 50 KJ/cm3 [21,30]. Radially
integrating the energy density over the ablated layer results in the ablated material fluence:
Φabl = Ldec ⋅ (1 − exp( Labl / Ldec )) , depicted in Fig. 2(e). The ablation layer length and fluence add
a nonlinear component to the calculation due to their threshold nature, and therefore retain the
figure of the absorbed fluence curve above a certain value.
The emitted plasma trajectory is calculated assuming a linear coupling coefficient
between the absorbed fluence in the ablated material and the resulting pressure, and the linear
trajectory of the ejected plasma normal to Al sphere. The average ablated material pressure at
a given angle on the sphere is: P R = C m ⋅ Φ abl , where the coupling coefficient, Cm, for Al equals
to 300 KN/J [23]. Based on geometrical scaling, the plasma pressure on the glass surface can
be written as P = P R ⋅ cos 2 θ , with the x-position on the surface given as x = R ⋅ tan θ . The
projected plasma pressure on the surface is shown in Fig. 2(e), showing a noticeable
‘shoulder’ very closely matching the observed depth profile of a wide LSP (see Fig. 1(a)).
The difference between the two polarizations and the smoothing of P polarization non-
monotonic profile could be explained as a result of the homogenization by the plasma
confinement effect [21].
3. Model validity
3.1 2D numerical calculation
Although a metal sphere on a planar silica glass surface illuminated by a linearly polarized
plane wave is a three-dimensional (3D) problem, it could be well represented by 2D
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9784

simulations using S and P polarization excitations as long as the sphere size is larger than the
wavelength. For spheres with a large enough radius of curvature (R » λ0) the deviations are
negligible between the 3D small out-of-plane surface slope and the 2D invariant geometry in
this direction. In that case, the 2D simulations with P and S polarization excitations represent
the cross-sections of the 3D simulations in the polarization plane and in the plane normal to it,
accordingly.
The results shown in Fig. 3 validate the 2D approach for particles of 1 μm radius and
above. For smaller particles of 0.2 μm the deviation between the 2D and 3D approaches
increases, yet provides qualitatively good predictions. Some of the apparent deviation in the
smaller particle size could be attributed to the observed ~0.4 nm increase in the decay length
for all angles with respect to the 2D case, which suggests that for these cases the 2D approach
could serve as a tight lower bound on the emitted plasma pressure. The 3D calculations were
conducted with a reduced maximal resolution of 10 nm, to accommodate the increased
computation expense. The validity of the 2D approach, which is substantially less
computationally intensive, enables us to continue this study with enhanced resolution, which
becomes especially challenging for the analysis of larger particles, and for cases where
symmetry considerations could not be applied (e.g., illumination at angle).

Fig. 3. 2D vs 3D modeling approaches results for an Al sphere on a glass exit surface excited
with Φin = 1 J/cm2. (a),(c) absorbed energy density fraction for R = 0.2 μm, and R = 2 μm,
accordingly, (b),(d) plasma pressure for R = 0.2 μm, and R = 2 μm, accordingly. The legend in
(b) holds also for all other sub-figures: 2D analysis for P (blue) and S (red) excitation, 3D
analysis results in the XY plane (magenta, dashed) and the ZY plane (cyan, dashed), both
excited in the XY plane.

3.2 The underlying physics of the model

The laser radiation interaction with the metal particle produces plasma, which in turn interacts
with the substrate underneath to result in a LSP. The processes associated with plasma
excitation and modes of confinement between the particle and the substrate were recently
discussed in [21]. In that work, the emphasis was on the description of the particle ejection
due to the interaction with laser radiation, whereas in this paper we focus on the effects
producing the substrate modification.
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9785

Plasma formation on the surface of a metal during exposure to a laser pulse has been
considered to be associated with the onset of intensive evaporation [27]. The onset of the
evaporation process occurs when the deposited energy density in the outer layer reaches a
threshold value of the enthalpy required to evaporate the material, Habl [21,31]. This
approximation holds as long as the energy is deposited in thin enough layer (~few nm) of the
material close to the surface. The optical absorption of the formed plasma is proportional to
its density which results in a rapid increase in the ablation and therefore in a sharp threshold
at Eabs = Habl and in the resulting relation Labl = Ldec ⋅ log( Eabs
R
/ H abl ) , which were confirmed for ns

pulses [32]. The local relation between the absorbed energy (fluence) in the metal and the
pressure of the ejected plasma follows the empirical findings of Phipps et al. for planar
geometries [27]. Here, we extend this correlation between absorbed fluence and emitted
plasma pressure for a non-planar geometry using experimental values for the conversion
coefficient [23] (which are also similar to the values found in [33]).
The ejected plasma is assumed to move with trajectories normal to the particle surface and
etch away the substrate surface on impact. Even though the etch process more directly relates
to the energy density of the plasma, we will present the calculated plasma pressure projected
onto the surface, and since the plasma pressure and energy density are monotonically related,
this pressure will indicate the etch-modified substrate geometry. This advancement with
respect to the scalar approach associating the incident fluence with plasma pressure, adds
geometry dependent details. As an example, for the values in the scenario detailed in section
3.1 and using the approach detailed here, we find that a small angular span (~200 mrad) under
the particle locally absorbs more (up to 12%) than away from a substrate (up to 8%), while
the angular side regions absorb less. The interference effect reduces the plasma generation
fluence threshold with respect to a scalar calculation, and also makes the plasma generated
more localized under the particle.
While including a rigorous full-wave approach for light interference in the structure, the
model takes a simplified approach towards the energy transport and plasma generation, based
on empirical observations. Since the thermal diffusion is neglected here, based on [32] for ns
pulses, for longer pulses including heat transport in the model could result in spatial smearing
of the heat absorption profile. Since this profile is much smaller than the scale of features of
interest in the azimuthal direction its effect is negligible with respect to the reported profiles.
However, since comparable to the decay length, we expect that the absorbed energy will be
lower and thus the plasma generation is lower than the model predicts. The assumption that
the plasma is ejected normal to the surface is a result of the plasma being generated in a small
layer on the surface, which still holds.
Recently [20] we reported in situ measurements of plasma emission associated with LSP
formation from steel spheres bound to fused silica surfaces. In that work, temperatures and
electron densities associated with ablated Fe atoms were observed to be in the 15,000-20,000
K and 3.5-4.4 × 1017 cm3 ranges, respectively, for LSP pit depths ranging from 0.5 to 1.2 μm.
As also described in the present work, the metal-rich plasma expands towards the substrate,
depositing a sufficient amount of heat into the silica to evaporate and ultimately ionize
substrate atoms. As the particle is propelled away from the substrate, the evaporated material
escapes and forms a crater. Based on the model presented here, plasma is created near the
point of contact with the particle, and a ‘shoulder’ appears in the pit profile which could be
traced back to the light absorption on the metal particle. A future improvement to this model
could be made through inclusion of the plasma confinement near the contact point [21]. In
[20], temperatures associated with etched Si atoms at the point of particle-substrate contact
were found to be ~2 × higher than that of the mostly unconfined Fe plasma indicating a local
plasma confinement [20]. The denser plasma in the confinement area near the contact point of
the particle and the glass surface, with higher heat capacity, may result in more absorption
and increased plasma temperature. This larger removal depth with respect to the model at that
region will result in a deepening of the bottom of the pit with respect to the model predictions.
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9786

An additional deviation from the model predictions, resulting from the high thermal
conductivity of the plasma, is the azimuthal homogenization of the plasma density profile
which could explain the more uniaxial shape observed in the measurements.
The model predictions for the ablation-driven plasma pressure can be compared to
confined plasma models for laser peening [31]. For the fused silica and Al interface, the
pressure in the plasma can be presented in terms of laser intensity as P = C1 ⋅ I , where C1 =
33 GW1/2/(cm Kbar). The predicted pressure for laser peening using the same parameters of
the scenario in section 3.1 (excitation fluence 1 J/cm2 and pulse length 1 ns) is about one
order of magnitude higher than the obtained values by our model. The expression for the laser
peening is expected to act as an upper bound, in agreement with these calculated values, since
it includes the plasma confinement effect. In addition, the model for laser peening also
predicts the amount of ablated fused silica Labl = C2 ⋅ P ⋅ Δt , where C2 = 0.013 μm/(ns Kbar).
Based on the pressure values obtained in this study, of about 10 Kbar, and for pulse durations
of nanoseconds, the predicted LSP depth is on the scale of few hundreds of nanometers,
which is consistent with the experimental findings [14,19,21].
The excitation source used in our numerical simulation is of a plane wave shape, while
physical sources have finite width. By comparing simulations excited by a plane wave source
with Gaussian beam sources of different waist-size, we find that the plane wave source well
represents Gaussian sources of waist larger than the particle radius (negligible changes are
observed when the beam waist equal the particle diameter). Therefore the plane wave
excitation used in the FDTD simulations represents well most of the finite size practical
excitation sources. When the Gaussian beam waist is comparable to the particle radius, the
illumination non-uniformity dominates the absorption profile and the side lobes in the
absorbed energy profile and the resulting signature ‘shoulders’ in the pressure profile are
found to be substantially reduced.
4. Parametric study
Next, we detail the results of a sensitivity study of the predicted emitted plasma profile for a
selected set of parameters: the excitation fluence, the particle size, the air-gap length between
the particle and the substrate surface, the thickness of an oxide layer surrounding the metal
particle (if one exists), and the input beam incidence angle.
4.1 Excitation fluence
The plasma generation, predicted by the model, increases with incident fluence, as illustrated
in Fig. 4. These findings agree with the experimental observation that the LSP depths increase
with incident fluence [19,21].
The source of the nonlinear plasma emission profile dependence on the incident fluence,
according to the model, is the thresholding nature of the required absorbed energy to cross the
ablation enthalpy. The FDTD numerical simulations are linear in intensity and therefore Ldec
R
is independent and Eabs is linearly dependent on incident fluence variations. The ablated layer
thickness grows logarithmically with the incident fluence as a result of the radial exponential
decay into the sphere, as shown for both polarizations in Figs. 4(a) and 4(c). The resulting
energy flux (fluence) contained in the ablated material and thus also the plasma pressure on
the glass surface increase sub-linearly with the input fluence, where at high input fluence the
energy densities are substantially larger than the ablation enthalpy and thus Φabl ≈ Φabs , taking
a close to linear dependence. The close to linear dependence on input fluence of the peak
plasma pressure, as well as the close to constant profile change, for high input fluences (>2
J/cm2) for both polarizations, is illustrated in Figs. 4(b) and 4(d).
At low intensities the plasma generation is located in the vicinity of the contact point.
With further incident fluence increase the plasma ejection area increases and due to the
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9787

interference mediated absorption distribution there is a sharp transition to the broader plasma
zone. The plasma ejection area in the 2D calculations is represented by the emission arc
angular length (i.e., the integrated angular arc length where ablation occurs) (see Fig. 4(e)). A
sharp transition is observed between a Gaussian shape for narrow pits, at low incident
fluence, and a ‘shoulder’ profile for wider pits, at higher incident fluence, in agreement with
experiment (as illustrated in Fig. 1(a)).

Fig. 4. Variation with excitation fluence. R = 1 μm Al sphere on a fused silica surface. S-

polarized excitation: (a) ablated layer thickness, (b) plasma pressure; P-polarized excitation:
(c) ablated layer thickness, (d) plasma pressure. The legend in (b) holds also for all other sub-
figures. The plotted plasma pressure is normalized by its peak value, P0, given in legend. (e)
The angular length of the emission arc vs the incident fluence, for both polarizations.

4.2 Particle size

The ‘shoulder’ deviation from the Gaussian-shaped plasma pressure on the surface, a
signature of the wave interference, is enhanced as the particle size increases, and is more
emphasized for P-polarized excitation. The transition from a close-to-Gaussian profile for
narrower pits to that with a ‘shoulder’ signature for wider pits resembles the trends observed
for LSPs [14,21]. For a 0.5 μm radius particle, the resulting plasma pressure profile has
insignificant ‘shoulder’ features as illustrated in Figs. 5(a) and 5(d), for two input fluences 1
J/cm2 and 4 J/cm2, accordingly. As the particle size grows the ‘shoulder’ signature grows as
well, as illustrated in Fig. 5.
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9788

The effective pressure area on the surface scales linearly with the particle size. The x-axis
in Fig. 5 is normalized by the particle radius, and the non-changing width of the plasma
profiles for changing particle size (for same Φin and polarization) is evident in Fig. 5.
Nevertheless, the plasma pressure peak power does not change substantially with the change
in particle size, and is insignificant with respect to the dependence observed on the incident
fluence (as detailed in section 4.1). In addition, the differences observed in Fig. 5 between the
two polarizations are reduced at the higher input fluence, which could be an additional
contributor to the LSPs observed axisymmetric shape.

Fig. 5. Variation in plasma pressure with particle sphere radius. Al sphere on glass surface at
Φin = 1 J/cm2 (a-c) and at Φin = 4 J/cm2 (d-f). S and P polarized excitation are compared for
different radius: (a,d) R = 0.5 μm; (b,e) R = 1 μm; (c,f) R = 2 μm. The plotted plasma pressure
is normalized by its peak value, P0, given in the legend.

4.3 Air gap between particle and surface

Varying the air gap difference, g, between the bottom of the metal particle and the glass
surface can be used to illustrate the interference effect in the presence of surface modification
(e.g., roughness). It allows us to gradually examine the change in the resulting plasma
pressure while transitioning between the two limit-cases: the particle on the substrate and the
particle suspended in air. This study could be applicable also with respect to studying plasma
nano-etching of glass by metal tips held in close proximity to glass surface and illuminated
from the substrate direction. In addition, these results could be instructive to the study of
particles of more complex geometries, which could be viewed as composed of multiple
spheres at different gap differences above the surface.
As the air gap increases from 0 to 1 μm, the absorbed power and thus also the emitted
plasma pressure asymptotically trend between the cases of a particle on a substrate and a
particle suspended in air, as illustrated in Fig. 6. The decay length into the metal was found
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9789

not to be significantly dependent on the air gap size and thus the absorbed fluence variation,
as depicted in Figs. 6(a) and 6(c) for both polarization excitations, is a result of the variation
of the surface energy density profiles. It is evident that for a particle located about 1 μm away
from the glass exit surface the resulting plasma pressure is close to Gaussian, without
significant side lobe modulations. It is also observed that the maximal peak pressure on
surface is obtained when the particle is on the surface.

Fig. 6. Variation with air gap between particle and glass surface (lifted particle). R = 1 μm Al
sphere, Φin = 1 J/cm2. (a-b) S-polarized, and (c-d) P-polarized excitations. (a,c) absorbed
fluence normalized by input fluence; (b,d) plasma pressure. Gap size, g, is given in legend: for
S polarization in (b), and for P-polarized in (d).

4.4 Oxide layer surrounding the sphere

The presence of oxide layer surrounding the metal sphere could result in an enhancement in
the absorbed power and thus in the emitted plasma pressure and also in a modulation of the
emission distribution. The formation of a few tens of nanometers oxide layer around metal
particles, and particularly alumina around Al particles, is a commonly occurring result of
exposure to air or humidity [34,35].
The effect of an alumina layer of thickness, t, on the plasma pressure distribution is
depicted in Fig. 7(a) and 7(b), for S and P polarization excitations, accordingly. The effect is
qualitatively similar for both polarizations: the plasma pressure emission is increased and the
distribution is reshaped with side-lobes as the layer increases towards 40 nm; with further
increase in the oxide layer thickness, the profile reduces back to a central lobe (i.e., maximal
emission directly under the particle) of similar size as the one without the oxide layer. This
follows the enhancement in the maximal absorbed energy density on the particle surface, as
illustrated in Fig. 7(c), suggesting that the enhancement around t = 40 nm is a result of
optimal oxide layer function as an impedance matching layer to the metal at that thickness.
This is further illustrated in Fig. 7(d) as the absorptivity is plotted as a function of the oxide
thickness. The absorptivity is calculated as the integrated absorbed fluence over the metal
sphere lower hemisphere (i.e., the absorbed energy in the particle) normalized by the
integrated incident fluence over the particle diameter (i.e., the energy incident on the sphere
foot-print). As a reference, the absorptivity is calculated for the particle suspended in air for
both the polarizations (dashed lines). The absorptivity of an oxide-free metal sphere is similar
with and without the interface, even though the locally absorbed fluence differs, as is apparent
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9790

from the results presented in section 4.3. Nevertheless, it is clear from Figs. 7(c) and 7(d) that
both the local absorption and the absorptivity of the sphere increase substantially due to an
oxidization layer (up to × 2 growth in local absorption, an up to × 1.5 – × 3.5 growth in
absorptivity depending on polarization). In Fig. 7(e), the spatial distribution of the E-field
intensification (intensity enhancement with respect to the excitation intensity) outside the
metal particle (i.e., in both oxide and air) is depicted for no oxide layer (t = 0), t = 40 nm, and
t = 80 nm. The resulting interference illustrates how the interference pattern is modulated,
building intensified side lobes near the metal surface at t = 40 nm. The effect of the
encapsulating oxide layer on the plasma emission is negligible since the oxide layer is only
few tens of nm. We note that the modified particle-substrate interface, being oxide-glass
rather than metal-glass, is expected to result in a lower Van der Waals attractive force (due to
about × 2.5 smaller Hamaker constant [36,37]), which would have some effect on the particle
detachment dynamics.

Fig. 7. Variation with alumina oxide layer thickness, t, surrounding the particle positioned on
the glass exit surface: R = 1 μm Al sphere, Φin = 1 J/cm2. Plasma pressure distribution for (a)
S-polarized and (b) P-polarized excitations. (c) Maximal absorbed energy density at the
particle surface, and (d) absorptivity as a function of t. (e) E-field intensities distributions
outside the metal for different t values (P-polarization).
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9791

4.5 Angle of incidence

Changing the light angle of incidence is of practical importance to optical systems designers
and is also illustrative to studying this effect since it modifies the symmetry of the
interference pattern. The model predicts that the resulting change in the interference pattern
generates a stronger plasma emission towards the direction of incoming light. The incidence
angle, α, with respect to the incoming beam from below the particle, is illustrated in Fig. 1(c).
The azimuthal distribution of the radial decay length is shifting towards the direction of
illumination, as illustrated in Figs. 8(a) and 8(d), for S and P polarizations, accordingly. In
addition, the varying gap between the bottom point of the sphere with respect to the incoming
beam and the glass flat surface result in a modified interference pattern, that is evident in the
energy density distribution on the sphere radius, depicted in Figs. 8(b) and 8(e). This results
in an asymmetrical and enhanced plasma pressure emission towards the illuminating
excitation, with apparent side lobes.

Fig. 8. Variation with excitation angle, α. R = 1 μm Al sphere, Φin = 1 J/cm2, on surface. (a-c)
S-polarized and (d-f) P-polarized excitations. (a,d) radial decay length, (b,e) absorbed energy
density on sphere surface, and (d,f) plasma pressure. α values are given in the legend for S
polarization in (c), and for P-polarized in (f).
 Vol. 25, No. 9 | 1 May 2017 | OPTICS EXPRESS 9792

5. Conclusion
We present here a model that predicts the laser-induced plasma emission from a surface-
bound micron-scale metal spherical particle, resulting in the modification by etching of the
glass substrate under the particle. The model starts with a full-wave electromagnetic
simulation of the geometry which results in the absorption distribution in the metal particle,
and using a simplified approach, which is justified by empirical findings, the plasma emitted
from the particle and thus the projected plasma pressure on the surface is calculated. The
model predications provide an explanation to the puzzling ‘shoulders’ signature observed
experimentally in the form of laser-induced shallow pits (LSPs), as well as agreeing with
other experimentally observed trends such as: the formation of the shoulders for wider LSPs;
increment of the LSPs size with the incident fluence; and correlation between the LSPs width
and depth. These findings highlights the significance of the interference of the incident light
in the open cavity geometry formed between the micro-particle and the substrate in the
resulting LSPs shape. It also provides further understanding for the dependence of the
resulting LSPs in the geometry found in experiments. For example: the LSPs profile shape
stays constant as its size grows with the incident fluence; the LSPs width is predicted to
increase linearly with the particle size; the emitted plasma pressure asymptotically trends
between the cases of a particle on a substrate (having side-lobes) and particle suspended in air
(without side lobes) as the air gap between the particle and the substrate is varied;
enhancement in the energy absorption in the metal particle substantially increases for specific
oxide layer thicknesses (40 nm in the studied case); and for non-normal incident light a
stronger plasma emission towards the direction of incoming light is observed. A better
understanding of the plasma emission from metal particles in the vicinity of a glass surface
could also be applied beyond the field of laser damage to potential applications in sub-
wavelength laser machining of transmissive materials through controlled plasma jet emission,
and for laser-target interaction in directed energy applications where the target is composed of
metal micro or nano structure on substrate.
Funding
Lawrence Livermore National Laboratory (LLNL) Laboratory Directed Research and
Development (LDRD) (14-ERD-098).
Acknowledgments
We would like to acknowledge R. N. Raman, S. G. Demos, R. A. Negres and I. L. Bass for
useful conversations about the modeling and the experimental observations of laser-induced
shallow pits and damage. This work was performed under the auspices of the U.S.
Department of Energy by Lawrence Livermore National Laboratory under contract DE-
AC52-07NA27344. LLNL-JRNL-723532-DRAFT.