The dependence of ion acceleration
on pulse length, pulse intensity,
and polarisation of a femtosecond
pulsed laser
by
Eric Leerssen
10188673Supervised by:
O. Versolato
W. Ubachs
R. Hoekstra
Second assessor:
E. van Heumen
University of Amsterdam Faculty of science Advanced Research Center for NanolithographyJuly 10, 2015
Abstract
The dependence of the ion acceleration on the pulse length, pulse energy, and polarisation of femtosecond laser pulses was studied. For the measurements a 800 nm Ti:Sapph laser with pulse lengths between 40 and 550 fs, pulse energies of 0.5 to 2.2 mJ and vertical and horizontal polarisation was used. Two different ion acceleration processes can be identified. One causing slow ions of ∼20 eV and one creating fast ions of 300 eV to several keV. The slow ions probably come from the main plasma. The fast ions could get there extra kinetic energy from ambipolar field acceleration. For shorter pulse lengths there is less ion production as well as lower ablation depth. Furthermore more fast ions can be detected perpendicular to the polarisation direction of the laser. The slow ions don’t show a significant change within scatter. This could be due to the imprinting of polarised laser light on the target surface.
Populair wetenschapelijke samenvatting
In de nanolithografie wordt er licht gebruikt voor het maken van struc-turen met een groote van enkele tientallen nanometers op silicium ”wafers”. Hoe kleiner de golf lengte van het gebruikte licht is, hoe kleiner de gemaakte structuren kunnen worden. Dit is een van de pro-cessen die Moores law, die zegt dat elke anderhalf jaar de snelheid van een chip verdubbelt, in stand houd. In de nieuwe generatie lithografie machines wordt licht van een golflengte van 13.5 nm gebruikt. De manier hoe ze dit licht maken is door een korte krachtige laser puls op een tin druppeltje te schieten. Dit tin druppeltje spat dan uit elkaar in een plasma gemaakt van tin ionen. Dit plasma straalt vooral veel extreem ultra violet (EUV) licht uit van 13.5 nm.
In de lithography industrie is effici¨entie erg belangrijk en men wil dan ook zoveel mogelijk EUV licht krijgen per hoeveelheid laser licht die voor de creatie nodig is. Dit wordt de conversie energie genoemd. Een manier die gevonden is om een hogere conversie energie te krijgen is een zogenaamd ”pre-pulse, main-pulse” systeem. Hierbij worden er in plaats van een krachtige laser puls die het tin druppeltje in een keer verdampt twee pulsen gebruikt. De eerste puls, de pre-pulse, is niet zo krachtig en zorgt ervoor dat het druppeltje van vorm veran-derd. Bijvoorbeeld in de vorm van een soort platte pannekoek. De tweede pulse is wel krachtig en kan omdat het druppeltje nu geen druppelvorm meer heeft deze veel effiecienter verdampen. Dit heeft er mee te maken dat het opervlak waarmee de laser het druppeltje raakt nu bijvoorbeeld veel groter is.
Meestal worden voor deze pre-pulses laser pulsjes met een lengte van nano secondes gebruikt. Het probleem hiermee is alleen dat er bij het maken van het plasma erg veel snelle ionen vrij komen. Deze ionen hebben zoveel energie dat ze potentieel de rest van de lithografie ma-chine kapot kunnen maken. Eerdere studies hebben aangetoond dat niet alleen nano seconde pulsen een tin druppel kunnen vervormen, maar dat femtoseconde (een femtoseconde is een miljoenste van een miljardste seconde) pulsen dit bijvoorbeeld ook kunnen. Aangezien deze puls zo kort is hebben alle atomen in het druppeltje geen tijd om te reageren voordat het pulsje alweer weg is en al zijn energie heeft afgedragen. Deze pulsen creeren dus op een andere manier plas-mas dan nanoseconde pulsen. Dit heeft als potentieel dat er een stuk minder snelle ionen worden afgestoten.
In deze scriptie wordt de afhankelijkheid van de ion versnelling van de pulse lengte, pulse energie en de polarisatie van de pulse bestudeert. Dit wordt gedaan door met femtoseconde pulsen op een stuk tin folie
af komen gemeten met detectoren onder verschillende hoeken.
Het blijkt dat er inderdaad bij korte femtoseconde pulsen een stuk minder snelle ionen worden geproduceerd. Waardoor dit komt is nog de vraag. Ook is het niet bekent of de tin druppels nog steeds even veel worden vervormd.
Al met al lijken femtoseconde laser pulsen een veelbelovend alter-natief voor het gebruik als pre-pulse, alleen wordt er fundamenteel nog niet goed begrepen hoe de ionen versneld wordt en meer onderzoek zal hier nog naar gedaan moeten worden.
Contents
1 Introduction 5
2 Theory 8
2.1 Femtosecond ablation . . . 8
2.2 Ambipolar field acceleration . . . 9
3 Experimental setup 10 3.1 Setup . . . 10 3.2 Ion detectors . . . 12 3.2.1 Faraday cups . . . 12 3.2.2 Channeltrons . . . 14 3.3 Data acquisition . . . 15 4 Measurements 16 4.1 Shot-to-shot fluctuation . . . 16
4.2 Consecutive shots on target . . . 16
4.3 Yield dependence on power . . . 20
4.3.1 Peak kinetic energy . . . 20
4.3.2 Peak yield . . . 21
4.3.3 Ablation depth . . . 22
4.3.4 Acceleration mechanisms . . . 22
4.4 Pulse length dependence . . . 25
4.5 Anisotropic effect in the ion emmision . . . 27
1
Introduction
In nanolithography light is used to write structures on silicon wafers. According to Rayleigh’s criterion the resolution of such a structure is of the order of the wavelength of the light. So to make ever smaller structures the wavelength of the light has to be decreased. The wave-length focused on by the next generation of nanolithography machines is 13.5 nm. One of the ways to produce this extreme ultraviolet (EUV) light is by creating a plasma of tin ions. These tin plasmas are created by shooting a pulsed laser on a tin droplet and thus ablating it into a plasma that emits strongly in the 13.5 nm range [1].
Since the industry wants to work as efficiently as possible the con-version factor of energy put into the laser versus the energy of the EUV light is very important. This is called the conversion energy. One of the ways to improve the conversion energy is by using a pre-pulse main-pre-pulse system. By first deforming the tin droplet with a low intensity laser pulse, the pre-pulse, and then hitting it with a far more powerful pulse, the main-pulse, the conversion energy has shown to increase significantly [2, 3].
When these low energy pre-pulses hit the tin droplet they can create high energy ions. These ions can damage the EUV collection system, and thus reduce machine life time. One possible solution might be the use of femtosecond pulses as a pre-pulse, instead of the nanosecond pulses that are usually used. The short interaction time of the laser pulse might be able to reduce the amount of high-energy ions created while still deforming the droplet. This deformation of a droplet, with the use of femtosecond pulses, can be seen in figure 1 where two droplets where hit with different laser parameters [4]. What they also have seen is that there is a dip in the amount of high energy ions produced when a 800 fs pulse length is used (figure 2).
This study aims to verify this result and furthermore research and fundamentally understand how the ion distribution reacts to different laser parameters. This is done by hitting a femtosecond pulsed laser on solid tin targets and changing the pulse lengths between 40 and 550 fs, pulse energies between 0.5, and 2.2 mJ and by varying the polarisation direction.
Figure 1: Droplet deformation using femtosecond laser pulses. Source: ISAN, EUV Litho Dublin.
Figure 2: Charge yield on a faraday cup at 5 and 8 keV against different pulse lengths, source: ISAN, EUV Litho Dublin.
2
Theory
In this section the theory behind the acceleration of ions with femtosec-ond pulses is explained. First the heating of the plasma is described and then a possible model for the creation of fast ions is presented.
2.1
Femtosecond ablation
When femtosecond pulses hit the tin target photons interact with the electrons via inverse bremsstrahlung. This process heats up the electrons. Since the timescale for energy transfer from the electrons to the ions is ∼1 ps during the time the laser hits the target there is no energy diffusion. After the laser pulse has gone the hot electrons can interact with the lattice and transfer their energy [5].
To ablate material from the surface with a femtosecond pulse the energy of the pulse has to be higher than the ablation threshold. The ablation threshold is determined by how the target absorbs the laser pulse and how much energy is needed to evaporate the material. The ablation threshold is given by:
Fth=
ρΩ
α (1)
where ρ is the density, Ω the specific heat of evaporation and α the material absorption coefficient which relates to the skin depth δ with α = 2δ.
For longer pulses, for instance in the nanosecond regime, the heat diffusion coefficient D has to be taken into account since a part of the energy deposited into the system diffuses through the lattice. The ablation threshold then becomes:
Fth = ρΩ
√
Dτ (2)
Where τ is the pules length. This makes the ablation threshold for nanosecond pulses much larger than for the femtosecond regime.
When the pulse fluence is higher than the ablation threshold the target can be ablated. This means ablation can take place up to a depth where:
Fa ' Fthexp(αz) (3)
with Fa the laser pulse fluence, Fth the ablation threshold, and z the
direction perpendicular to the surface.
The expected temperature of the lattice and the expected ablation depth is then:
T ' Faα
L ' α−1ln(Fa Fth
) (5)
Where T is the temperature, C the heat capacitance, and L the ablation depth.
If we assume that al the outgoing ions are distributed with the Maxwell speed distribution we would expect from equation 4 that the energy of the peak scales linearly with the laser fluence.
If we assume that all the ablated material is distributed isotropi-cally we expect the ion yield at a specific opening angle to scale with the laser fluence logarithmically as seen in equation 5 since a higher laser fluence means a bigger ablation depth and more material de-posited.
2.2
Ambipolar field acceleration
An acceleration mechanism that could bring the ions to higher energy is time-dependent ambipolar field acceleration [6]. The laser creates a space-charge layer of electrons that create a time-dependent ambipolar field. Since this layer consists of negatively charged electrons it pulls the positively charged ions out of the lattice. This accelerates the ions to higher energies than they would get with a thermal mechanism.
When the field that is created by the space-charge is assumed to look like a plate capacitance the increase in kinetic energy that the ions get can be calculated with the following formula:
∆Ekin=
Ze2NeλD
0S
(6) Where Ze is the ionic charge, e is the electron charge, Ne is the
amount of electrons, 0 is the free-space permittivity, and S is the
surface of the field (in this case the laser spot size). The Debye length λDis the measure of how far the net effects of a certain charge carrier
travel and is given by:
λD= s 0kBTe q2 ene (7) Where 0and kBare the vacuum permittivity and Boltzmann
con-stant respectively. Tethe electron temperature, qethe electron charge,
3
Experimental setup
In this section the setup used in the experiments is described. In the following paragraphs more will be explained about the laser, vacuum system, target and detectors.
3.1
Setup
The laser that is used in the experiments is a pulsed Ti:sapphire laser. It has an output wavelength of 800 nm, an output frequency of 1 kHz, produces pulses with a length between 40 fs and 2 ps, and has a pulse energy up to 3 mJ.
Figure 3: Schematic drawing of the experimental setup. To have more control over the output frequency of the pulses a Galvo mirror is used for pulse picking. It picks a pulse by flipping a mirror very fast between pulses. The Galvo mirror is used at a 5 Hz frequency with a 0.5% duty cycle, effectively sending pulses at only 5 Hz into the system. The rest of the pulses are dumped on a beam dump. A lower output frequency gives two benefits. The first being that there is a larger response time to start the measurement and the second that the tin foil target needs some time, typically of the order of 100 ms, to move in between shots.
To measure the length of the used pulse an auto-correlator is used. The energy of the pulse is measured with a power meter and the polarisation is checked by minimising, or maximising, the reflection on a thin film polariser by tuning a λ2 wave plate.
Figure 4: The vacuum system.
The laser is focused with a lens that has a focal length of 1000 mm and then enters the vacuum system through a BK7 non-AR coated viewport. The vacuum system can be seen in figure 4.
The target that is used is a 1 mm thick 99.999% pure tin foil. It is moved with the use of two PI Q-545 stages which are controlled by PI E-871 controllers. These stages have a micrometer precision in positioning which gives us sufficient control over the tin target.
The system is under vacuum so that ions do not react with air. When ions move through air they can have elastic and inelastic colli-sions with the atoms in the air. These collicolli-sions can cause the ions to be deflected or neutralised resulting in them not being detected. To negate these effects the system is pumped down to a vacuum between 1 × 10−6mbar and 1 × 10−8mbar with the use of two turbo-molecular pumps. This significantly increases the mean free path of the ions and the charge exchange even for highly charged ions is negligible.
To trigger the whole system the triggering signal of the laser is used. This signal is sent to a Stanford Research Systems DG645 delay generator. The delay generator then sends a signal to the Galvo mirror to send a laser pulse into the system, to the PI controllers to move the stage, and to the Keysight Infiniium oscilloscope to measure the signals from the ion detectors.
3.2
Ion detectors
In this experiment two different kinds of ion detectors are used: Fara-day cups and Channeltrons. The FaraFara-day cups are used for measuring the total ion charge that comes out of the plasma. The Channeltrons have a large gain factor, with which they can measure small signals. The downside to this is that without exact knowledge of the gain factor the total charge information is lost.
The coordinate system that is used for the positioning of the de-tectors is a spherical coordinate system. The (φ, θ) = (0, 0) axis corre-sponds with the laser propagation as can be seen in figure 5. A Fara-day cup in the horizontal plane, 30 degrees from the laser propagation would then be called FC(30,0). In the measurements three Faraday cups are used: two in the horizontal plane FC(2,0), FC(-30,0) and one in the vertical plane FC(0,30). Furthermore there is one Channeltron in the horizontal plane used in the system called CHT(30,0).
x z y Laser FC(φ, θ) −φ θ
Figure 5: The coordinate system for the Faraday cups.
3.2.1 Faraday cups
A Faraday cup is a device that measures ions by the charge that they deposit on it. The Faraday cup mainly consists of three elements as can be seen in figure 6, the cup, the suppressor, and the shielding. The cup (number 1) is the part that measures the ion signal. When a charged particle hits the cup it transfers its charge. The current
Figure 6: On the left: cutthrough of a Faraday cup with 1: the cup, 2: the suppressor and 3: the shielding. On the right: a picture of one of the used Faraday cups.
between the cup and ground is measured over a 10 kΩ resistance with the oscilloscope. The charge that hits the cup can be calculated with C = It where C is the charge that hits the cup, I the current, and t the bin time of the scope.
When an ion hits the cup there is a possibility to create secondary electrons. This negative charge leaving the cup also creates a current which can not be distinguished from the actual ion signal and thus making it seem that there are more ions hitting the cup than there actually are. For this reason the suppressor (number 2 in figure 6) is used. The suppressor is put on a small negative bias of −20 V so that any secondary electrons created on the cup get pushed back onto the surface of the cup.
To make sure that only ions coming directly from the plasma can hit the Faraday cup and to prevent the suppressor potential to interact with the ions before they enter the cup the front shield (number 3 in figure 6) is added. The aperture in the shielding has a diameter of 6 mm. With that the opening angle can be calculated to compare the different detectors at different positions. The solid angle is given by: Ω = πrR22 Where Ω is the solid angle in steradians, r is the radius of
the aperture of the detector and R is the distance from the tin foil target to the detector. For the FC(-30,0) and FC(0,30) R =25 cm and r =3 mm giving a solid angle of 4.5 × 10−3sr.
is added to prevent the incoming beam of ions from reflecting of the vacuum vessel wall onto the Faraday cup. Also an aluminium cone is added reaching from the shield towards the disc. These adaptations to the design make sure that no ions can go around the shielding and hit the Faraday cup from behind.
Electrons are also produced during the ablation process. These electrons are not a real problem since even electrons with small ki-netic energies, of only a couple of eV, can reach significant speeds. The electrons reach the cup long before the ions do and this signal can easily be recognized. A bigger problem form the secondary elec-trons created by ions hitting surfaces close to the Faraday cup. These secondary electrons have the same time of arrival as the ions but with there negative charge cause a lower signal. The negative bias of the suppressor is also used to prevent these electrons from hitting the cup. Tin ions can be in a lot of different charge states. Since the Faraday cup only measures the charge that is deposited on them and not the amount of ions the difference between one 10+ tin ion and ten 1+ tin ions can not be measured. This can make the results hard to interpret.
3.2.2 Channeltrons
Channeltrons are channel electron multiplier detectors. When a charged particle hits the front of the channel, which is shaped like a funnel, it creates secondary electrons as seen in figure 7. Those secondary electrons are then pushed further into the tube by a large electric field from the negative bias on the front of the channel. These accelerated electrons then hit the channel again creating more electrons, thus cre-ating a very large signal with only a small input. At the end of the channel an anode is situated to measure the amount of electrons. More can be found about Channeltrons in [7].
When the rate at which the secondary electrons are created is higher than the rate at which they can be replaced by the bias cur-rent, supplied by the high voltage supply, the Channeltron can get saturated. When this happens the relation between the Channeltron input and the current measured at the anode is not linear and the data interpretation is hindered. The output current should typically be limited to below 10-20% of the bias current or else the Channel-trons get saturated. Since we are working with a bias voltage of 950V and the Channeltrons are measured to have a resistance of 60 MΩ the bias current is around 15 µA. This leaves us with a maximum output current of 1.5 µA. A measuring resistance of 10 kΩ is chosen to oper-ate the Channeltrons with a good resolution without being saturoper-ated. With this resistance the maximum voltage without the Channeltrons being saturated is 15 mV. The downside to this is that this resistance
Figure 7: An ion hitting the channeltron and causing a shower of secondary electrons. Source: Channeltron handbook [7].
also brings a non negligible RC-time with it typically of ∼ 1 µs given R = 10 kΩ and C ≈ 125 pF
When a low signal is expected, for instance on the large (∼60◦) an-gles with respect to the laser propagation, Channeltrons can provide a significant benefit over the Faraday cup. Since all measurements in this theses were done at 30◦and 2◦, where there is sufficient signal for the Faraday cups, Channeltron data is not included. Future ex-periments that include detectors at 60◦, or other positions where low signal is expected, could well benefit from a Channeltron.
3.3
Data acquisition
For the data acquisition a Keysight Infiniium DSO9063A oscilloscope is used. It measures 9000 channels over a timespan of 0.5 ms giving a 50 ns temporal resolution. The signal is transferred from the Faraday cup with the use of a 1 m long BNC cable. For the measurements a variable resistance (Thorlabs VT1), which is set at a resistance of 10 kΩ, is used. The BNC cable and Faraday cup have a measured resistance of 124 pF. This gives an RC-time of 1.24µs.
4
Measurements
In the experiments laser pulses with a pulse length varying between 40 and 550 fs were shot on a tin foil target. By shooting on the target tin is ablated and a plasma is created. With the use of Faraday cups the ions originating of this plasma are detected in time-of-flight (TOF) traces.
In the following paragraphs the results of the measurements are shown. In the first paragraph the shot-to-shot fluctuations are ex-plained. Then the power and pulse length dependence are shown. In the last paragraph the change of the TOF trace due to the changing polarisation of the laser light is shown.
4.1
Shot-to-shot fluctuation
The first thing that stands out from the data is that for different positions on the tin target, but for the same amount of shots, the time of flight traces change as can be seen in figure 8, where the eight shot on the target can be seen for four different positions on the target. An explanation for this could be that the laser has a fluctuation in output power, but this is measured to be less than 1% of the output power. More likely is that this is caused by the fact that there are irregularities on the tin foil surface. There are many scratches on the surface of the foil. Furthermore the oxide layer on the foil is also not perfectly smooth.
Because of this scattering it is difficult to see a significant change in the data. To get a clearer look at the effects we assume that the scatter is normally distributed and take the average over 100 shots. In figure 9 this average is given along with the standard deviation.
4.2
Consecutive shots on target
Multiple shots are fired at one single position on the target. There is a difference in the time of flight traces from shot-to-shot that can not just be explained by scatter (figure 10). Three different peak struc-tures can be distinguished, which are indicated in the figure. The am-plitude of the first peak decreases sharply with the number of pulses, with saturation after 4 shots. The amplitude of the second fast peak does not seem to change as much from shot to shot. The third peak however changes significantly from shot to shot. The amplitude of this peak is stable after 10 shots.
Energy dispersive X-ray (EDX) measurements have been performed on the tin foil, with special thanks to Nick Spook and the ARCNL
0 30 60 0,000 0,005 0,010 Yield(V) Time(µs)
Figure 8: The eighth shot on the target on four different spatial locations.
EUV targets group for doing the measurements. These EDX mea-surements show a difference in the oxygen and carbon content of a piece of ablated (figure 11 spectrum 3) and non-ablated tin (figure 11 spectrum 2). Spectrum 1 is of a piece of tin that is hit by the wings of the gaussian laser pulse. It is clearly seen that the oxygen and carbon content of the spectrum that was hit with the focus is less than the spectrum that was untouched. The oxygen and carbon level of the spectrum that was hit by the wings is between that of the shot and not-shot regions.
So more shots, or more intensity in a region means less oxygen and carbon in that region. This leads us to believe that this first fast peak in figure 10 consists of oxygen and carbon ions. These ions would come from a layer of oxide that is ablated by the first five shots. This has also been observed in previous research [8].
Assuming they are oxygen ions the peak energy of the first fast peak is around 200 eV. The second peak could be associated with fast tin ions which would have a kinetic energy of around 1 keV. The third peak probably consists of slow tin ions with a peak kinetic energy of 20 eV. This same slow and fast peak distribution of ions has also been seen in previous research [6].
Figure 9: Average over 100 shots, the shaded reagon gives the 1 sigma con-fidence band.
Figure 10: Fluctuation of TOF trace for different amount of shots, three different peaks are indicated. Two fast, one slow.
Figure 11: Energy dispersive X-ray measurement of the tin surface with spectrum 2 of a non-shot surface, spectrum 1 of a position where the wings of the gaussian laser beam hit the target and spectrum 3 in the middle of the laser focus.
Figure 12: TOF trace for different pulse energies for the FC(0,30)
4.3
Yield dependence on power
The measurements shown in this paragraph present the difference in TOF trace for different pulse energies (figure 12). These measure-ments were done with a pulse length of several hundred femtoseconds. The same three peak structure as in figure 10 can be seen again. The amplitude of the first fast peak, associated with oxygen, is very small and can be safely ignored. The second and third peak can be asso-ciated with Snx+. In figure 13 the kinetic energy distribution for the first 5 measurements are shown. The fast tin ion peak for the 2.14 mJ trace has a kinetic energy of around 1 keV and the slow ion peak 20 eV.
4.3.1 Peak kinetic energy
In the top graph of figure 14 the kinetic energies of these fast and slow peaks measured on the FC(0,30) and FC(-30,0) Faraday cups are given for different pulse energies. This is measured with a horizontal polarisation. To compare the fast and slow peak these values are normed against the 2.14 mJ signal. Both the fast and the slow peaks kinetic energy show a linear dependence on pulse energy. But the fast peak seems to have a stronger dependence than the slow peak.
From equation 4 we would expect that for a higher laser fluence the temperature of the lattice also increases. From the Maxwell-Boltzmann speed distribution we know that the median speed of
par-Figure 13: dQ/dE trace for different pulse energies for the FC(-30,0)
ticles coming out of a hot gas scale with the square root of the tem-perature of the gas. From this follows a linear relation between the pulse energy and the peak kinetic energy. The data of figure 14 shows this nicely.
4.3.2 Peak yield
The change of the yield of the peak for different pulse energies is shown in the bottom graph of figure 14 again normed against the 2.14 mJ trace. The fast and slow peak both scale with pulse energy. There seems to be an exponential scaling with pulse energy of the slow peak.
Equation 5 says that the ablation depth scales logarithmically with the pulse energy. When we assume an isotropic distribution of ions this would mean that the total ion yield would also scale logarithmically. But from figure 14 it can be seen that the yield of the slow peak scales exponentially with pulse energy. This difference could be explained by the fact that the Faraday cup measures charge and not ions and that with higher pulse energies more highly charged ions are created.
4.3.3 Ablation depth
The ablation depth is measured with a high NA optical microscope. This is done by focusing the microscope at the edge of the crater and in the middle of it. The difference in focal depth between the two positions is the ablation depth for 30 shots on the target.
From figure 15 it is clear that the ablation depth indeed has a logarithmic scaling with the pulse energy agreeing with equation 5.
4.3.4 Acceleration mechanisms
From the information of paragraph 4.3.1 together with the fact that the energies of the slow ions are around 20 eV it can be assumed that ions in the slow ion peak are from the main plasma.
Although the fast ion peak shows the same scaling on pulse energy as the slow peak, the dependence on pulse energy is much stronger for the fast ions. Furthermore, the kinetic energies of these ions (around 1 keV) are too high to be produced in a thermal mechanism (a 10 million Kelvin plasma would be needed).
Time-dependent ambipolar field acceleration could be an expla-nation for these high energies. The temperature associated with the plasma that created the slow ion peak is around 230 × 103K. With an assumed electron density of 2.545 × 10−4cm2 and taking the number of electrons to be Ne≈ 5 × 1011, which are the common values found
in literature for these intensities, this would give us a Debye length of λD≈ 10 nm. With an area of the beam spot of 2.545 × 10−4cm2, and
assuming Z = 1, we can calculate the added kinetic energy a Sn+ion gets from this ambipolar field using equation 6. Which gives us an added kinetic energy of a few keV. Within an order of magnitude this explains the energy of the fast tin ion peak. This model could also explain why the kinetic energy of the fast peak shows a far greater dependence on the pulse energy. When the energy of the the pulse is reduced the temperature of the lattice goes down reducing λD but
0 . 0 0 . 2 0 . 4 0 . 6 0 . 8 1 . 0 1 . 2 1 . 6 2 . 4 3 . 2 4 . 0 4 . 8 5 . 6 6 . 4 7 . 2 8 . 0 8 . 8 F l u e n c e ( J / c m ^ 2 ) p e a k e n e rg y n o rm a liz e d [ a .u .] f a s t p e a k f c ( 0 , 3 0 ) s l o w p e a k f c ( 0 , 3 0 ) f a s t p e a k f c ( - 3 0 , 0 ) s l o w p e a k f c ( - 3 0 , 0 ) 0 . 4 0 . 6 0 . 8 1 . 0 1 . 2 1 . 4 1 . 6 1 . 8 2 . 0 2 . 2 - 0 . 2 0 . 0 0 . 2 0 . 4 0 . 6 0 . 8 1 . 0 1 . 2 p e a k y ie ld n o rm a liz e d [ a .u .] P u l s e e n e r g y ( m J ) f a s t p e a k f c ( 0 , 3 0 ) s l o w p e a k f c ( 0 , 3 0 ) f a s t p e a k f c ( - 3 0 , 0 ) s l o w p e a k f c ( - 3 0 , 0 )
Figure 14: Top: The kinetic energy of the fast and slow ion peak for different pulse energies. Bottom: The yield of the peak of the fast and slow ion peak for different pules energies. The traces are normed on the 2.2 mJ trace. The polarisation is horizontal. Pulse length of order of few 100 fs.
1 0 0 1 0 2 0 3 0 4 0 5 0 A b la ti o n d e p th ( µ m ) L a s e r p u l s e p e a k f l u e n c e ( J / c m 2)
Figure 15: The ablation depth against the laser fluence. The three colors are three different series of measurements.
4.4
Pulse length dependence
Another important laser parameter is the length of the pulse. The pulse length was varied from shorter pulses below 100 fs to longer pulses of several 100 fs. All measurements were done with a pulse energy of 2.05 mJ and a vertical polarisation.
For the shorter pulse lengths there seems to be less ion yield then for the longer pulses. The fast and slow peak seem to have the same
dependence on pulse energy. This can be seen in figure 16 where
the yield is given for increasing pulse length. The dashed line is the supposed symmetry axis and the shortest pulse length. The decrease in ion yield for the use of shorter pulses seems significant, it is more than a factor ten less than the yield of the longer pulses.
The kinetic energies of the peaks show a slightly different behaviour (figure 17). Although the kinetic energy of the fast peak shows the same strong scaling with pulse length, the short peak only shows an increase of ∼10%.
Again ablation depth measurements have been done (figure 18). Just like the ion yield the ablation depth shows a strong dependence on the pulse length. This means that not only are there less ions detected, that total mass that is ablated is also reduced by changing the pulse length.
Figure 16: Relative peak yield at different pulse lengths for the slow and fast peak. The dashed line is the supposed symmetry axis and the shortest pulselength.
Figure 17: Peak Kinetic energy at different pulse lengths for the slow and fast peak. The dashed line is the supposed symmetry axis and the shortest pulselength.
Figure 18: The ablation depth at the center of the crater for different pulse lengths. The different colors stand for different series of measurements. The dashed line is the supposed symmetry axis and the shortest pulselength.
4.5
Anisotropic effect in the ion emmision
When the polarisation of the laser beam is changed it seems to have an effect on the distribution of the ions. In figure 20 can be seen that when the polarisation is changed from 0 to 90 degrees with respect to the horizontal plane the amplitude of the fast peak increases and decreases. The largest yield is there when the polarisation direction is perpendicular to the plane where the detector is in (for instance a vertical polarisation for the horizontal Faraday cup). The yield for the 45◦polarisation is exactly in the middle of the 0 and 90 degree polarisations for both detectors. In the bottom graph of figure 20 the horizontal and vertical traces of both the FC(-30,0) and FC(0,30) are combined. It is interesting to see that the traces with different po-larisations almost overlap, making this an almost perfectly symmetric effect. There is no significant effect visible on the slow ion peak.
From research in surface science it has shown that polarised laser light can imprint anisotropic surface structures [9, 10]. When the next laser pulse hits these structures they could cause the anistropic distribution of ions. From electron microscope pictures made from the target it can be seen that there are some sort of anisotropic structures on the surface (figure 19) potentially causing this ion distribution.
Figure 19: Electron microscope pictures of a region ablated by the laser. Vertical wave like structures can be seen.
0 3 0 6 0 0 . 0 0 0 0 . 0 0 4 0 . 0 0 8 0 . 0 1 2 Y ie ld ( V ) Time (µs) 90° polarisation fc(-30,0) 90° polarisation fc(0,30) 0° polarisation fc(-30,0) 0° polarisation fc(0,30) 0 . 0 0 0 0 . 0 0 4 0 . 0 0 8 0 . 0 1 2 Y ie ld ( V ) 90° polarisation fc(0,30) 45° polarisation fc(0,30) 0° polarisation fc(0,30) 0 . 0 0 0 0 . 0 0 4 0 . 0 0 8 0 . 0 1 2 0 3 0 6 0 Y ie ld ( V ) 90° polarisation fc(-30,0) 45° polarisation fc(-30,0) 0° polarisation fc(-30,0)
Figure 20: Top figure: the different polarisation on the FC(-30,0) detector. Middle figure: the different polarisation on the FC(0,30) detector. Bottom
5
Conclusion
The measurements done with different laser parameters contain a lot of information about the possible ion acceleration mechanisms. The two peak structure in the tin ion traces point towards two different ion acceleration mechanisms. The slower ions with low kinetic energy come from the main plasma. The fast ions have too much energy to be accelerated this way and could get their extra energy from ambipolar field acceleration. For further investigation it would be interesting to know the charge states of the ions as well as the electron distribution coming of the plasma since these are import parameters for ambipolar field acceleration that are not yet known.
When the pulse length is varied there seems to be a significant change in the ion yield. At the lowest pulse length there is a decrease in ion signal of around a factor ten compared to the longer pulse lengths used. Also the kinetic energy of the fast peak shows a dependence on pulse length, but the slow ions only show a small dependence. So not only is the total amount of fast ions less, they are also less energetic. This is the exact opposite as found in previous research and it would be interesting to measure the pulse length behaviour for different intensities, to see how this dip behaviour depends on the laser intensity and maybe get some more information about the mechanisms causing this decrease in yield. Furthermore it would also be interesting to see how this femtosecond pre-pulse interacts with a tin droplet at these pulse lengths to see if such a pulse is still able to deform a droplet.
Only the fast ion peak seems to depend on the laser polarisation. There are more ions going perpendicular to the polarisation direction then parallel. A possible explanation for this is that the polarised laser light imprints an anisotropic surface structure on the tin foil, but more research has to be done to confirm this model.
Concluding, interesting physical effects can be seen with femtosec-ond pulsed laser produced plasmas. The data has shown us that by adjusting the length and polarisation of the laser pulse you can have a significant measure of control over the direction, energy, and amount of ions.
References
[1] V. Bakshi, EUV Sources for lithography. SPIE, bellingham, WA, 2005.
hancement of kev x-ray emission in laser-produced plasmas by a weak prepulse laser,” Applied Physics Letters, 1987.
[3] S. Fujioka, M. shimomura, Y. Shimada, S. Maeda, H. Sakaguchi, Y. Nakai, and T. Aota, “Pure-tin microdroplets irradiated with double laser pulses for efficient and minimum-mass extreme ul-traviolet light source production,” Applied Physics letters, 2008. [4] A. Vinokhodov, V. Krivtsun, M. Krivokorytov, Y. Sidelnikov,
S. Chekalin, V. Kompanets, A. Melnikov, and K. Koshelev, “Femtosecond laser pre-pulse technology for lpp euv source.” http://www.euvlitho.com/2014/S72.pdf. Accessed: 08-july-2015. [5] B. Chichkov, C. Momma, S. Nolte, F. von Alvensleben, and A. Tunnermann, “Femtosecond,picosecond and nanosecond laser ablation of solids,” Applied physics A, 1996.
[6] S. Amoruso, X. Wang, C. Altucci, C. de Lisio, M. Armenante, R. Bruzzese, N. Spinelli, and R. Velotta, “Double-peak distribu-tion of electron and ion emission profile during femtosecond laser ablation of metals,” Applied Surface Science, 2002.
[7] Photonis, Channeltron electron multiplier handbook for mass spectrometry applications.
[8] N. Farid, S. Harilal, H. Ding, and A. hassanein, “Kinetics of ion and prompt electron emission from laser-produced plasma,” Physics of plasmas, 2013.
[9] J. V. Obona, V. Ocelik, J. Skolski, V. Mitko, G. Romer, A. H. in’t Veld, and J. D. Hosson, “On the surface topography of ultrashort laser pulse treated steel surfaces,” Applied Surface Science, 2011. [10] J. Z. P. Skolski, G. R. B. E. Romer, A. H. in’t Veld, J. Obona, V. Ocelik, and J. D. Hosson, “Modeling of laser induced periodic surface structures,” Journal of Laser Micro/Nanoengineering, 2010.
