MASTER'S THESIS
Development of Laser-Induced
Breakdown Spectroscopy for Analyzing
Rinds and Layered Structures in Martian
Rocks
Love Alm
Master of Science in Engineering Technology
Space Engineering
Development of laser-induced breakdown
spectroscopy for analyzing rinds and layered
structures in Martian rocks
Love Alm
Lule˚
a University of Technology
Department of Space Science
Abstract
Laser-induced breakdown spectroscopy (LIBS) offerers the possibility to study rinds and layered structures in a sample by creating a crater by employing repeated laser ablation. Previous studies have shown that the existence of craters can cause significant changes in both the thermal properties and the composition of the plasma.
The occurrence of contributions to the plasma thorough plasma-crater interactions was studied for Martian atmospheric conditions using a sam-ple consisting of a dolomite slab covered by an aluminum foil. The ex-periment showed that significant amounts of aluminum was present in the plasma after the laser had penetrated the foil completely. In four craters of different depth and a diameter of 1400 µm the aluminum content of the plasma was significantly lower when using a 600 µm beam diameter for spectral acquisition as compared to when using a 1400 µm beam. Ad-ditionally the differential penetration due to the Gaussian shape of the beam intensity profile was found to be a complicating factor by reducing the depth resolution of the LIBS measurement.
The change in the composition of the plasma due to elemental frac-tionation was studied by observing the intensity of selected Al, Ca, Fe, K, Mg, Na and Si emission lines throughout crater formation. A homo-geneous basalt slab was analyzed under Martian atmospheric conditions. Only Mg and Na exhibited a change in line intensity throughout the crater formation that is large enough to indicate a change in the composition of the plasma. When acquiring spectra using a beam with a diameter of 600 µm in craters with a diameter of 1400 µm the rate of change in line in-tensity for the different elements were not significantly different from the reference values. The conclusion is that using different two beam diam-eter, a larger for crater formation and a smaller for spectral acquisition improves the ability to separate between different layers in the sample without increasing the problem of changes in fractionation.
Acknowledgments
While the thesis in itself represents the final six months of my studies. The results are the product of a five year journey both at the university and outside. Completing my master’s thesis and concluding my studies would not have been possible without the help of...
...my supervisor professor Seiji Sugita for giving me the opportunity to work in his group at University of Tokyo. Your help and encouragement helped me push though the most intensive periods.
...all the members of professor Sugita’s group, for welcoming me to the group and giving me much valuable advice.
...my family, for their never ending support in helping me realizing my goals.
...Lack of Sanity (LoST) and the hang-arounds, my coterie of trust-worthy characters in the darkest corner of the room.
...my classmates in the space engineering program. When worked to-gether not even the sky could limit us.
...the people Kashiwa Lodge. We rode out earthquakes, tsunamis and the threat of nuclear melt-down together.
...Sweden-Japan foundation for helping me to finance my stay in Japan. Your support made my stay possible and memorable.
Contents
1 Introduction 13 2 Literature review 15 2.1 Fundamentals of LIBS . . . 15 2.2 Ambient atmosphere . . . 17 2.3 Back condensation . . . 17 2.4 Crater interactions . . . 18 2.5 Elemental fractionation . . . 19 3 Methodology 21 3.1 Objectives . . . 21 3.2 Equipment . . . 223.2.1 Nd:YAG laser - Continuum Surelite I . . . 22
3.2.2 Spectrograph - Ocean Optics USB4000 . . . 22
3.2.3 Imaging optics . . . 23
3.2.4 Sample chamber . . . 23
3.3 Spectral acquisition . . . 24
3.4 Baseline subtraction and normalization . . . 24
4 Experiment 1 - Investigation of rind contributions to the plasma during crater formation 29 4.1 Objectives . . . 29 4.2 Execution . . . 29 4.3 Results . . . 30 4.3.1 Aluminum . . . 31 4.3.2 Calcium . . . 33 4.3.3 Magnesium . . . 33 4.4 Conclusions . . . 36
5 Experiment 2 - Influence of beam diameter on rind contribu-tions to the plasma 39 5.1 Objectives . . . 39 5.2 Execution . . . 39 5.3 Results . . . 41 5.3.1 Aluminum . . . 41 5.3.2 Calcium . . . 45 5.3.3 Magnesium . . . 45 5.4 Conclusions . . . 50
6 Experiment 3 - Investigation of elemental fractionation during crater formation 53 6.1 Objectives . . . 53 6.2 Execution . . . 53 6.3 Results . . . 54 6.3.1 Aluminum . . . 54 6.3.2 Calcium . . . 54 6.3.3 Iron . . . 54 6.3.4 Magnesium . . . 56
6.3.5 Potassium . . . 56
6.3.6 Silicon . . . 58
6.3.7 Sodium . . . 58
6.4 Conclusions . . . 60
7 Experiment 4 - Influence of beam diameter on elemental frac-tionation 63 7.1 Objectives . . . 63 7.2 Execution . . . 63 7.3 Results . . . 65 7.3.1 Aluminum . . . 65 7.3.2 Calcium . . . 65 7.3.3 Iron . . . 65 7.3.4 Magnesium . . . 69 7.3.5 Potassium . . . 69 7.3.6 Sodium . . . 73 7.3.7 Silicon . . . 73 7.4 Conclusions . . . 77
8 Discussion and future work 79
List of Figures
3.1 Overview of the equipment setup . . . 22
3.2 Example spectra and baseline from a dolomite sample . . . 27
3.3 Example spectra and baseline from a basalt sample . . . 27
4.1 Evolution of total spectral intensity . . . 30
4.2 Comparison of the evolution of the aluminum spectral lines . . . 31
4.3 Evolution of Al II spectral line at 281.61 nm . . . 32
4.4 Evolution of Al II spectral line at 308.82 nm . . . 32
4.5 Evolution of Al II spectral line at 623.17 nm . . . 32
4.6 Comparison of the evolution of the calcium spectral lines . . . . 34
4.7 Evolution of Ca II spectral line at 317.93 nm . . . 34
4.8 Evolution of Ca II spectral line at 393.37 nm . . . 34
4.9 Comparison of the evolution of the magnesium spectral lines . . 35
4.10 Evolution of Mg I spectral line at 383.23 nm . . . 35
4.11 Evolution of Mg I spectral line at 517.27 nm . . . 35
4.12 Comparison of the evolution of selected aluminum, calcium and magnesium spectra lines . . . 37
4.13 Al II, 623 nm at conclusion experiment 1 . . . 37
5.1 Comparison of the Al II spectral line at 281.61 nm acquired using 600 µm and 1000 µm beam diameter . . . 42
5.2 Comparison of the Al I spectral line at 308.82 nm acquired using 600 µm and 1000 µm beam diameter . . . 43
5.3 Comparison of the Al II spectral line at 623.17 nm acquired using 600 µm and 1000 µm beam diameter . . . 44
5.4 Comparison of the Ca II spectral line at 317.93 nm acquired using 600 µm and 1000 µm beam diameter . . . 46
5.5 Comparison of the Ca II spectral line at 393.37 nm acquired using 600 µm and 1000 µm beam diameter . . . 47
5.6 Comparison of the Mg I spectral line at 383.23 nm acquired using 600 µm and 1000 µm beam diameter . . . 48
5.7 Comparison of the Mg I spectral line at 517.27 nm acquired using 600 µm and 1000 µm beam diameter . . . 49
5.8 Al II, 623 nm at conclusion experiment 2 . . . 50
6.1 Evolution of total spectral intensity . . . 54
6.2 Evolution of Al I spectral line at 308.82 nm . . . 55
6.3 Evolution of Ca II spectral line at 393.37 nm . . . 55
6.4 Evolution of Fe I spectral line at 374.95 nm . . . 55
6.5 Evolution of Mg I spectral line at 517.27 nm . . . 57
6.6 Evolution of K I spectral line at 766.49 nm . . . 57
6.7 Evolution of K I spectral line at 769.89 nm . . . 57
6.8 Evolution of Si I spectral line at 288.21 nm . . . 59
6.9 Evolution of Na I spectral line at 588.99 nm . . . 59
6.10 Evolution of Na I spectral line at 589.89 nm . . . 59
7.1 Comparison of the Al I spectral line at 308.82 nm acquired using 600 µm and 1000 µm beam diameter . . . 66
7.2 Comparison of the Ca II spectral line at 393.37 nm acquired using 600 µm and 1000 µm beam diameter . . . 67
7.3 Comparison of the Fe I spectral line at 374.95 nm acquired using 600 µm and 1000 µm beam diameter . . . 68
7.4 Comparison of the Mg I spectral line at 517.27 nm acquired using 600 µm and 1000 µm beam diameter . . . 70 7.5 Comparison of the K I spectral line at 766.49 nm acquired using
600 µm and 1000 µm beam diameter . . . 71 7.6 Comparison of the K I spectral line at 769.89 nm acquired using
600 µm and 1000 µm beam diameter . . . 72 7.7 Comparison of the Na I spectral line at 588.99 nm acquired using
600 µm and 1000 µm beam diameter . . . 74 7.8 Comparison of the Na I spectral line at 589.89 nm acquired using
600 µm and 1000 µm beam diameter . . . 75 7.9 Comparison of the Si I spectral line at 288.21 nm acquired using
List of Tables
3.1 Summary of Surelite I . . . 25
3.2 Summary of USB4000 . . . 25
3.3 Result of sample chamber leakage test . . . 25
4.1 Beam configuration used in experiment 1 . . . 30
4.2 Spectral lines used in experiment 1 . . . 30
5.1 Craters parameters used in experiment 2 . . . 40
5.2 Beam configuration used in experiment 2 . . . 40
5.3 Spectral lines used in experiment 2 . . . 40
6.1 Beam configuration used in experiment 3 . . . 53
6.2 Spectral lines used in experimet 3 . . . 53
6.3 Summary of regression parameters . . . 61
7.1 Summary of craters used in experiment 4 . . . 64
7.2 Beam configuration used in experiment 4 . . . 64
7.3 Spectral lines used in experiment 4 . . . 64
7.4 Comparison of the slopes of the regression lines for the spectral lines in experiment 3 and 4 . . . 78
1
Introduction
An important part in understanding the origin and geology of Mars is to study its petrology. On NASA’s Viking, Pathfinder and Mars Exploration Rover mis-sions this was achieved by the use of Alpha-particle x-ray spectrometer (APXS). While APXS offers high accuracy and low detection levels, the method is not ideal for Martian applications. The APXS instrument must be in direct con-tact with the sample and the measurement time is in the range of a few hours. With the limited mobility of the previous Mars rovers this imposed severe lim-itations on the number of samples that could be studied. LIBS analysis on the other hand, can be performed at stand-off distances over ten meters and can be performed in a matter of minutes. This allows for a more rapid data acqui-sition from a higher number of samples. LIBS also makes the rover capable of studying samples in otherwise inaccessible locations. This should result in an overall higher scientific output of the mission. LIBS analysis is generally con-sidered a non-destructive since the amount of material removed by each shot is very small, often in the nanogram range and has a penetration depth in the micrometer range. With repeated laser pulses significant depth penetration can be achieved by employing the laser a a drill. This allows LIBS instruments to remove surface dust [1], rinds and creating depth profiles in the samples. These advantages have been acknowledged by NASA which have included the LIBS instrument ChemCam on their latest rover Mars Science Laboratory (MSL,) which is due for launch in November 2011[2]. This is an important step since it will be the first time that a LIBS instrument will be used on another planet.
The original objective of this thesis is to investigate complications that may arise when studying rinds and other layered structures in Martian rocks, devise and finally test the new methods. The topic as initially stated is not specific enough to be completely covered in the time available for the thesis. For this reason a preliminary investigation of the subject was conducted. The goal was to identify factors which had the potential to greatly improve quantitative analysis of layered samples using LIBS and that are relevant for Martian conditions to be studied further.
2
Literature review
2.1
Fundamentals of LIBS
There are substantial differences between different implementations of laser-induced breakdown spectroscopy (LIBS.) They can consist of two main com-ponents; a laser for inducing a plasma and a spectrograph for analyzing the line emissions of the plasma. There are several different types of lasers are commonly used, operating at wavelengths from infra-red to ultraviolet. Simi-larly great differences in pulse length, beam profile and irradiance can be found between different models.
The purpose of the laser is to create a vapor or plasma consisting of the sur-face material of the target. The plasma can be generated according to several different ablation mechanism which affect the nature of the plasma. For lasers operating with femtosecond pulses most of the ablation is of a non-thermal na-ture due to the shot duration and high irradiance. These mechanisms include; absorption by free electrons, multi-photon absorption, tunneling and avalanche ionization. For high energies, multiple charged ions can be produced through Coulomb explosions. In this case the pulse duration is shorter than the average reaction time for the ablation mechanisms. Since the plasma is formed after the pulse has ended the laser does not interact with the plasma. For picosec-ond laser pulses the ablation may be of both thermal and non-thermal nature depending on the irradiance of the beam. Electrons ejected from the target sur-face can absorb photons gaining energy causing ionization of the surrounding air. When employing nanosecond pulses of low irradiance most of the abla-tion is due to thermal mechanisms. This imposes a greater dependence on the thermal properties of the target. For irradiances in excess of 0.1 GW/cm2,
non-thermal mechanism may contribute to the ablation process. For nanosecond lasers the pulse duration is long enough for the laser pulse to interact and be partially absorbed by the induced plasma This phenomena is known as plasma shielding[3].
The induced plasma is a mixture of ions, electrons, neutral atoms, molecules and in some cases nanoparticles. In the early, hot plasma, the spectrum is domi-nated by continuum emissions from Compton scattering and re-combination. As the plasma expands into the surrounding medium, the temperature and number density decreases and the atomic emission lines begins to appear. Two processes accounts for the discrete line emissions; de-excitation of ions and de-excitations of neutral atoms. The wavelength of the spectral lines corresponds to the energy release when the atom moves between tow quantum states. The intensity of the spectral peak corresponds to the probability for that transition to occur and the total number of atoms of the element[3].
Qualitative LIBS analysis of a sample is a fairly straight forward process. The presence of a clear spectral line which is unique for a certain element is enough to prove the presence of the element. Quantitative LIBS analysis is on the other hand considerably more complicated. There exists several different mechanism which changes the composition of the plasma as well as the appear-ance of the spectrum. Since different materials have different thermal properties such as temperature of evaporation, there is no guarantee that the composition the plasma is identical to the composition of the sample. To further complicate matters the plasma interacts with the emitted light through various processes,
such as self absorption and line broadening which changes the appearance of the spectrum. Under the assumption of local thermal equilibrium (LTE) the line-to-continuum ratio can be calculated using the following equation[3];
l c = CrA21 1 + 4Ni++ Ni+ g2 Zion λc2 λlTe
expEion−E2
kBTe h ξ1 − exp−hνc kBTe + G · exp−hνc kBTe i (2.1)
G is the free-free Gaunt factor, A21Einsteins transition probability, g2 the
de-generacy parameter, Zion the ion partition function, λc the continuum center
wave length, λl the emission line center wave length, νc is the continuum
fre-quency, ξ the free-bound continuum correction factor and Cr is a constant of
2.005 · 10−5 sK.[3] The ratio Ni++
Ni+ can be determined using the Saha equation
[3]. Ni++ Ni+ =(2πmekBTe) 3/2 h3 2Z++(T ) Z+(T ) 1 Ne exp −Eion ++ kBTe (2.2) Eion++ is second ionization potential, Tethe electron temperature, nethe
elec-tron number density, kB is Boltzmann’s constant, h is Planck’s constant, Z+
and Z++ the partition functions of singly and doubly ionized atoms. Using
equation 2.1 and equation 2.2 we can determine the plasma temperature using the spectrum from the plasma.
In what is known as ’Stark broadening’; collisions in the plasma induces small changes in the wavelength of emitted photons thus broadening the spectral peak.
∆λ1/2= 2W " Ne 1016 1 + 1.75A Ne 1016 1/4 1 − 3 4N −1/3 D !# , (2.3)
where NDis the number of particle in the Debye sphere, W the electron impact
parameter and A the ion impact parameter. The polynomial approximation for A and W can be found in Ref [4]. Due to the line widening the changes in spectral peak height are not directly proportional to the abundance of the element. It is from this equation possible to determine the electron density of the plasma which is useful in plasma diagnostics. A commonly used method to measure the plasma temperature is to study the line-to-continuum ratio of the emitted light.
There are two general approaches to LIBS quantitative analysis, calibration free LIBS (CF-LIBS) and statistical analysis. CF-LIBS employs models for vari-ous physical mechanisms which influences resulting the spectrum. This includes the interaction the laser’s interaction with the sample, different ablation mech-anism and how the emitted light interacts with the plasma. This in an attempt to compensate for the distortion of the spectrum induced by various matrix effects and plasma interactions. In order to achieved accurate predictions of the samples composition CF-LIBS requires sophisticated models and accurate plasma diagnostics. The second approach is performed by creating a statistical training model using LIBS spectra from samples of known composition. The composition of the unknown samples are then predicted by comparing the spec-tra to those of the samples included in the spec-training model. Several different approaches has been used such as partial least square (PLS), principal compo-nents analysis (PCA) and soft independent modeling of class analogy (SIMCA)
each with their own advantages and disadvantages. The drawback of statistical analysis is that the accuracy of the training model in predicting the composi-tion of the unknown samples greatly depends on the types of sample included in the training model. In order to achieve accurate predictions the training model must contain samples of a similar composition as the unknown samples[5]. The statistical approach is therefore less flexible but often simpler to implement.
2.2
Ambient atmosphere
Traditionally LIBS has been performed on earth, using normal atmospheric pressure and composition. Earth’s atmosphere consists of 78.08 % N2, 20.95
% O2 (by volume) and trace amounts of Ar, CO2 Ne, He, CH4, Kr, and H2.
The pressure at sea level is 760 torr.[6] Mars exhibits both signifcantly lower pressure and a different composition. The Martian atmosphere at mean radius is characterized by 95.32 % CO2, 2.7 % N2, 1.6 % Ar, 0.13 % O2, 0.08 % CO
(by volume) and trace amounts H2O, NO, Ne, HDO, Kr and Xe with a total
pressure 5.8 torr.[7]
In an article from 2000; penetration depth, mass ablation and spectral evo-lution of the plasma was investigated using a 1064 nm Nd:YAG laser irradiating a 6061 aluminum alloy target in a CO2 atmosphere. The pressure ranged from
0.01 torr to 590 torr.[8] Between 0.01 and 100 torr only small variations in pen-etration depth was found. The highest penpen-etration depth occurred around 10 torr. The mass ablation rate increased with decreasing pressure but leveled out at pressures less than 1 torr. The explanation given was that the lower ambient pressure allows the plasma to expand more rapidly therefore greatly reducing the attenuating effect from plasma shielding. In a 100 % argon atmosphere the intensity of the 777.19 nm oxygen line decreased with a factor 2 and the previ-ously strong carbon line at 247.86 nm could not be detected. This indicates that the CO2 atmosphere of Mars can make significant contributions to the carbon
and oxygen content of the plasma[8].
Since the atmospheric pressure dictates the expansion rate and thus cooling rate of the plasma it also affects the spectral evolution. The early, hot plasma exhibits strong continuum emission while the cool, expanded plasma shows a higher degree of line emissions. For this reason LIBS spectra from normal atmosphere are often collected using a time gated spectrograph in order to achieve a higher line-to-continuum ratio. In the case of low ambient pressure the rapid expansion of the plasma quickly results in reduced continuum emissions. Previous LIBS experiments under Martian atmospheric conditions have achieved high line-to-continuum ratios using ungated spectrographs[9].
The findings indicate that the atmospheric conditions, both pressure and composition, can have a great influence on the LIBS measurements. This in-dicates that in order for the results to be useful for Martian applications the experiments must be conduced under similar atmospheric conditions.
2.3
Back condensation
Back condensation is dependent on several factors including temperature of condensation of the element, which in turn is closely related to the ambient pressure. For these reasons condensation tends to be a preferential process,
favoring some elements over others. In an article from 2007 the authors mod-eled the condensation processes of a laser-induced plasma in vacuum [10]. The article takes into account several different processes; laser interaction with the surface, evaporation and back condensation, transfer and absorption of the laser in the plasma, gas-dynamic expansion, heating and ionization of the plasma and generation and transfer of thermal radiation. With a single 20 ns (FWHM) shot from a 1064 nm, Nd:YAG laser at a peak irradiance or 0.5 GW/cm2
conden-sation rather the vaporization was found to be the dominating process during the second half of the pulse duration. The explanation is that during the first half of the pulse an optically thick plasma is formed which can result in up to 97 % attenuation of the beam. This results in a rapid decrease in surface tem-perature which combined with a high vapor pressure leads to conditions allows condensation to occur on the irradiated surface[10].
As condensation depends on the ambient atmosphere the results from Ref. [10] can not be directly transfered to Martian conditions. As discussed in Section 2.2 and Ref. [8] plasma shielding appears to be of little concern and showing only small variations in the pressure range found on Mars.
2.4
Crater interactions
Two separate models for how crater formation changes the conditions for LIBS measurements will be discussed, laser-crater interactions and plasma-crater in-teractions. Laser-crater interactions are difficult to measure accurately but the fundamental concepts has been studied previously by using a radiative transfer model for laser irradiation[11]. The difference in energy transfer between a flat surface and a cratered surface was estimated by comparing how much energy could escape the surface in the form of either black body radiation or through reflection. The study assumed an non-participating plasma and is thus only applicable as long as the induced plasma can be considered optical thin. The results from the study shows two important factors in determining the extent of the craters influence, the heat transfer to the surface and the depth / diameter aspect ratio of the crater. For a surface with very high heat transfer ratio only small differences was found between a flat surfaces and a crater. This can be explained by the fact that it is not possible to attain a heat transfer exceeding a 100 % of the beam energy. In the case of surfaces with low energy transfer parameter a dramatic change in energy absorbed is predicted during crater for-mation. The greatest changes in energy transfer to the sample was found in the early stages of crater formation[11].
Not only the laser’s interactions with the target changes due to crater forma-tion. The laser-induced plasma can similarly interact with the crater changing its properties. This phenomena was studied using a 266 nm Nd:YAG with 3 ns pulses having an irradiance of 7 GW/cm2. The sample was fused silica
and the experiment was conducted under normal earth atmospheric conditions. The laser was fired into pre-machined craters with the depth of 480 µm and a crater aspect ratio of 0, 1, 3 and 6. Both plasma temperature and electron number density increased with increasing crater aspect ratio. This effect was most pronounced while plasma was inside the crater. Both plasma temperature and electron number density decreased as the plasma expanded towards the crater rim but was still elevated when the plasma had expanded well outside the crater. Plasma-crater interactions also influenced the line intensity of the
288.21 nm silicon peak. For the plasma generated in the crater the silicon line’s intensity was more the three time stronger then for the plasma generated on a flat surface.
A subsequent study of plasma temperature and electron number density as a function of irradiance found a clear discontinuity in both when the irradiance reached the threshold irradiance of particulate ejection. The study was con-ducted using a quadrupled, 266 nm, Nd:YAG laser on a fused silica target at normal atmospheric pressure and composition. When exceeding the threshold irradiance a rapid increase in both plasma temperature and electron number density occurred as well as a discontinuity in the rate of change. The thresh-old irradiance depends on the target material but for the fused silica target the threshold irradiance was strongly linked to the crater’s aspect ratio. Three cases were studied, a flat surface, a crater with aspect ratio 2 and a crater with aspect ratio 4 was studied. The corresponding threshold irradiance decreased from 20 GW/cm2 to 9 GW/cm2 and finally 5 GW/cm2[4].
The possibility of the crater wall contributing to the plasma have been stud-ied using a Nd:YAG 1064 nm laser, operated with 6 ns (FWMH), 135 mJ pulse energy, an irradiance of 7.9 GW/cm2 and a repetition rate of 10 Hz. The
am-bient atmosphere was of normal Earth pressure and composition. Two set of measurements were made. In the first measurement an aluminum sample was irradiated directly. As expected, the spectra revealed strong aluminum peaks. In the second measurement the aluminum sample was covered by a pierced brass disk which effectively formed a crater surrounding the target area. The aperture of the brass disk was large enough so that the laser could irradiate the sample without coming into direct contact with the brass disk. The spectra revealed aluminum as well as copper and tin, which were not present during the first measurement[12]. This is a clear indication that the induced plasma can interact with the crater walls and in the process contribute material to the plasma.
2.5
Elemental fractionation
Elemental fractionation is a collective name for processes which causes the in-duced plasma to have non-stoichiometric properties. Examples of elemental fractionation include preferential vaporisation and condensation. The influence adds an additional complicating factor when performing quantitative analysis and efforts should be made to investigate and minimize its effects.
In an article from 1999 the influence of crater formation on fractionation was studied. Three different lasers were used; a 1064 nm Nd:YAG laser, a 266 nm Nd:YAG laser and a 248 nm excimer laser. The experiment was conducted using two different glass reference materials; NIST SRM 610 and BCR 126A in an ambient atmosphere of either 100 % helium or 100 % argon at normal pressure. The composition of the plasma was determined using inductively coupled plasma mass spectrometry (IPC-MS). The degree of fractionation was defined by the following equation:
φat =
Rr0/Rrt
Ra0/Rat
; (2.4)
where Ra0 is the initial signal response for element a, Rr0 the initial signal
t and Rrt the signal response for the reference element a at time t. Yttrium
was used as the reference element. The experiment showed that elemental frac-tionation increased in both samples with an increasing crater aspect ratio. The degree of fractionation showed large variations between the different elements in the samples. Calcium showed very little fractionation even for the maximum aspect ratio of approximately 10 in both samples. In contrast, zink exhibited a maximum fractionation of approximately 3.5 at the greatest aspect ratios. In general fractionation was small for crater aspect ratios less then 6. The authors lists Zn, As, Se, Cd, Te, Tl, Pb and Bi as the elements that showed the greatest increase in fractionation with increasing crater aspect ratio. When lowering the beam irradiance from the maximum of 10 GW/cm2to 7.5 GW/cm25 GW/cm2 and finally 2.5 GW/cm2the fractionation increased. The largest increase could be seen for the volatile elements Zn, Pb, Cr, Tl, Bi and Sb. Furthermore the crater profile changed drastically from a near cylindrical crater for the highest irradiance to a narrowing, needle like crater at an irradiance of 2.5 GW/cm2.
The explanation given by the authors is that larger parts of the beam having an irradiance which is lower than the ablation threshold results in a low de-gree of ablation outside the central region of the beam. The authors conclude that a high irradiance is necessary to minimize fractionation. The difference in fractionation between helium and argon atmosphere was studied finding that a 100 % helium atmosphere offered lower fractionation as a function of the crater aspect ratio compared to a 100 % argon atmosphere[13].
The ratio between U238 and Pb208 in NIST SRM 610 as a function of the
crater aspect ratio was studied in an article from 2000[14]. The reason for choos-ing U238and Pb208 is that they have a very different melting temperatures. A
266 nm quadrupled Nd:YAG laser was to produce a plasma and the composition of the plasma was analyzed using IPC-MS. As indicated in Ref. [13] the shape of the crater became increasingly needle shaped with repeated ablations. There-fore not only the crater bottom is irradiated but also significant proportions of the crater walls. The cross sectional irradiance remained constant but the the actual irradiance of the irradiated surface decreases for needle like craters. The study found that for an actual irradiance above 0.3 GW/cm2 the U
238/Pb208
ratio in the plasma was stoichiometric. At lower irradiances Pb208 was over
3
Methodology
3.1
Objectives
In a homogeneous target contributions from the crater walls are of no con-sequence since they do not affect the composition of the plasma. In layered samples such as rocks with rinds, this implies that the rind could contribute to the plasma after it has been penetrated by the laser. Previous studies have shown that rind contributions can have a significant impact on the composition of the plasma[12]. The experiment was however conducted using normal earth atmosphere and may not applicable to Martian atmospheric conditions.
A constant degree of fractionation can be compensated for during the LIBS data analysis. Changes in the degree of fractionation can however not be com-pensated for in the same manner without a priori knowledge of the sample. Fractionation have been shown to be dependent on the crater aspect ratio and thus the crater depth[14]. Since crater formation is necessary in order to study samples with rinds the degree of fractionation can be expected to change to some degree during the experiments. This results in three objectives for the thesis.
1. Investigate if crater interactions can affect the ability to study rinds and layered structures in rock samples.
2. Investigate if fractionation can affect affect the ability to study rinds and layered structures rock samples.
3. Investigate methods to improve the ability to study rinds and layered structures rock samples.
In order to complete the three objectives five tasks must be completed. 1. Build and configure an experimental set up using pre-existing components. 2. Plan experiments in accordance with the objectives.
3. Conduct experiments. 4. Perform data analysis.
3.2
Equipment
The experimental equipment consists of four main components; a Nd:YAG laser with focusing optics, imaging optics, a sample chamber capable of producing Martian conditions and a spectrograph. An overview of the equipment setup can be found in Figure 3.1.
Figure 3.1: Overview of the equipment setup
3.2.1 Nd:YAG laser - Continuum Surelite I
LIBS can be performed with a wide range of lasers each with their own ad-vantages and disadad-vantages. In this experiment a Continuum Surelite I, 1064 nm Q-switched Nd:YAG laser was used. The Q-switch is a standard method for controlling the pulse length and pulse energy distribution by denying or allowing oscillation in the optical cavity. By controlling the delay between flashlamp fir-ing and signal to open the optical cavity the length of the pulse can be controlled. By only allowing oscillation when the flashlamp’s output is the strongest the laser pulse intensity becomes more homogeneous[15]. A summary of the laser can be found in Table 3.1
The beam diameter at the target surface is controlled using a single convex lens with a focal length of 500 mm. The lens is mounted on an optics rail which allows for a lens-to-sample distance of 370 mm to 500 mm. With this configuration the lower limit of the beam diameter is the diffraction limit. The highest achievable beam diameter is 1.43 mm.
3.2.2 Spectrograph - Ocean Optics USB4000
Ocean Optics’ integrated spectrography system USB4000 was used for the ex-periments . The spectrograph employs a linear CCD array in order to record the spectra. The calibration data of the spectrograph is stored on an internal EEPROM memory allowing the spectrograph to be moved between computers with out the requirement of conducting additional calibrations[16]. The spec-trograph is controlled specspec-trography computer software OPWave though a USB
interface. The software is designed for controlling the spectrograph and acquire data and is equipped with limited post processing abilities[17].
3.2.3 Imaging optics
In order to utilize the full spectral range of the spectrograph some additional considerations has to be made. Normal glass has a low UV transmittance and would decrease the sensitivity significantly in this important band. It is there-fore important to use imaging optics made from a material which has a high UV transmittance, such as quartz. The simplest set up for imaging the light emitted from the plasma is by employing two convex lenses. This configuration is however susceptible to chromatic aberrations since the lenses will have differ-ent focal length for differdiffer-ent wavelengths. This results in a detection bias where certain wavelengths will be suppressed. This leads to an instrument dependent detection bias which should be avoided, With no achromatic, UV capable lens systems available the light emitted by the plasma was collected by a optic fiber cable mounted coaxially to the laser beam with its aperture 300 mm from the sample. This approach while possible in the laboratory is not suited for filed applications. The configuration will however allow the spectrograph to operate in its full spectral range.
3.2.4 Sample chamber
The sample chamber is equipped with two view ports for irradiation and imag-ing. One of the view ports was equipped with a quartz window in order to allow UV-light to be transmitted. This view port was subsequently used for both irradiation and imaging. The sample chamber is mounted on a stand which is capable of vertical translation. This allows for irradiation on a new area of the sample without opening the chamber. The chamber is connected to an manifold which in turn connects to a vacuum pump and to one or more bottles of gas. The chamber can be evacuate to approximately 0.1 torr and then re-pressurized with either air or gas from one of the bottles in order to control the atmospheric composition.
In order to determine the leakage rate of the sample chamber the cham-ber was evacuated and the isolated from the vacuum pump using one of the manifolds valves. The change in pressure inside the chamber was subsequently recorded. The results can be found in Table 3.3. After 45 minutes the test was aborted since the leakage was small enough that no changes in pressure could be seen over the last 15 minutes. Since the pressure during the experiments will be in the range of 7-10 torr, a full order of magnitude higher then the pressure at the time the leakage test was aborted there is no reason to assume that the leakage should cause significant changes in the atmosphere composition or pres-sure over the course of the experiments. The test chamber is determined to be suitable for further experiments.
The sample chamber requires manual control to maintain the correct envi-ronment. This is achieved by first evacuating the chamber to the limit of the vacuum pump and then re-pressurizing it with the desired gases. The cham-ber is then evacuated until the desired pressure is reached and is isolated from the vacuum pump using the manifold’s valve. One cycle of evacuating and re-pressurizing to normal atmospheric pressure using CO2 results in an air/CO2
ratio of approximately 0.001. For lower ratios it is possible to employ repeated cycles of evacuation and re-pressurization using the desired gas mixture. For the experiments two repeated evacuation cycles were employed resulting in an atmosphere consisting of <99.99 % CO2at a pressure of 7 torr.
3.3
Spectral acquisition
The spectrograph was configured to continuously acquire spectra with an inte-gration time of 100 ms. With the laser set to continuous irradiation at 10 Hz each integration cycle of the spectrometer will contain one laser pulse. While the laser pulse may occur at any point of the integration cycle no spectral data will be lost due to the continuous spectral acquisition. If the laser pulse occurs very late in the integration cycle most of the line emissions from the plasma may occur in the next integration cycle. This should however be of little con-cern since each spectra will contain the full evolution of the plasma and the difference in emission between subsequent shots is assumed to be small. The spectra from the first shot may however in some cases not contain all of the line emissions which can affect the subsequent analysis. The first spectra from the experiments will therefore not be included in the analysis.
During crater formation the lens-to-sample distance will change as a result of mass ablation. This will lead to a change in beam diameter and subsequently beam irradiance. Compensating for the change in lens-to-sample distance would either require beforehand knowledge of the penetration depth, which is sample dependent or the ability to measure the crater depth during the experiment. For this experiment neither option is feasible. The lens used to focus the laser beam has a focal length of 500 mm and the lens-to-sample distance is in the range of 370 to 500 mm. Since the crater depth is expected to never exceed a few millimeters this should have a negligible impact the lens-to-sample distance and thus the beam diameter and irradiance. The lens-to-sample distance will therefore be allowed to increase us the crater is formed.
3.4
Baseline subtraction and normalization
The first step in the analysis of the acquired LIBS spectra is baseline subtraction. Baseline subtraction allows for better comparison between different spectra and increases the dynamic range of the spectral lines. There are several different approaches to baseline subtraction. In these experiments locally weighted scat-ter plot smoothing (LOWESS), was employed. The algorithm assumes that the spectral peaks are noise (Ei) on an otherwise smooth curve g(xi)
Yi= g(xi) + Ei, (3.1)
where g(xi) is a linear approximation
g(xi, ~θ) = θ0+ xiθ2. (3.2)
The initial regression is performed using: ˆ θ(xo) = arg min n X i=1 K(xi− xo h ) {yi− [θ0− θ1(xi+ xo)]} 2 (3.3)
Table 3.1: Summary of Surelite I
Parameter Value
Wavelength (fundamental) 1064 nm
Pules length 5 ns
Pulse energy (all wavelengths) 445 mJ
Min repetition rate 0.5 Hz
Max repetition rate 20 Hz
Beam spatial profile (% fit to gaussian) Near field (< 1 m) 0.7
Far field (∞) 0.95
Table 3.2: Summary of USB4000
Parameter Value
Focal length at input 42 mm Focal length at output 68 mm
Optical design Asymmetrical crossed Czerny-Turner
Spectral range 200 - 850 nm
Spectral resolution 1.6 nm (FWHM)
Detector Toshiba TCD1304AP Linear CCD array
Detector dimensions 3648 linear elements
Pixel size 8µm x 200 µm
Signal-to-noise ratio 300:1
Interfaces USB and RS-232
Table 3.3: Result of sample chamber leakage test
Pressure (Torr) Time (minutes)
0.1 0 0.15 1 0.2 4 0.25 15 0.3 30 0.3 45
K is the so called tri-cube kernel weight function: K(u) =max 1 − u3 , 0 3 (3.4) which is zero outside the range of h as defined in equation 3.5. This results in that only the points that are located in the interval x0− h < xi < x0+ h are
used in the local regression. The value of h determines the number of points that are used in the local regression and thus the degree of smoothing from the regression. Determining the correct value of h is critical for accurately fitting the baseline. In general large values of h will under estimate the baseline while the opposite is true for small values of h. The regression is then refined using the robust regression estimator:
ˆ θ(xo) = arg min n X i=1 wr(xi)K( xi− xo h ) {yi− [θ0− θ1(xi+ xo)]} 2 (3.5) The factor wr is Turkeys bi-square weights which defined by;
wr(xi) =
n
maxh1 − (ri/b)2, 0
io2
(3.6) The weights determines the robustness of the regression, in other words how strongly outliers will influence the results. The last stage is to determine the scale parameter σ which is estimated by:
ˆ
σM AV = meridian (|yi− ˆgi(xi)|) /0.6745 (3.7)
The LOWESS regression is performed iteratively by computing g(xi) using
equa-tion 3.3. The robust weight and scale parameter is calculated using equaequa-tions 3.6 and 3.7, respectively. A robust regression is performed using equation 3.5 after which the kernel weights are calculated using equation 3.4. The last two steps are then repeated until the solution converges[18].
The LOWESS baseline estimation was implemented using the software pack-age ’R’ using the built in function ’lowess’. Trial and error showed that setting h to 10 % of the number of spectrograph channels and completeing 20 iterations resulted in a good fit for the baseline. Additional iterations did not alter the re-gression line significantly but increased computing time. Some of the data points of the estimated baseline were higher then the actuall baseline of the spectra. This typically occurred around sharp depressions in the spectra. In such cases the data points from the regression was replaced by their corresponding data points in the spectra. This ensures that no negative intensities are present in the baseline subtracted spectra. Negative intensities can interfere with the nor-malization of the spectra. Example of spectra and calculated baseline from two different rock sample during different stages of crater formation can be found in Figures 3.2 and 3.3. The LOWESS regression line is then subtracted from the spectra in order to produce a baseline subtracted spectra.
Before extracting the line intensity, the spectra was first normalized to have a total intensity of 1,000,000 a.u. This ensures that an overall decrease or increase in spectral intensity is not interpreted as a change in the composition of the plasma. Changes in the total spectral intensity could be due to many different factors which does not influence the composition of the plasma such as; to shot-to-shot variations in the laser beam, changes in sample distance, the ability of the optics to collect the emitted light or plasma confinement in the crater.
Figure 3.2: Example spectra (black) and baseline calculated using LOWESS (red) from dolomite sample. The figure on the left is the spectra from the first laser shot and the right from the 101:th laser shot.
Figure 3.3: Example spectra (black) and baseline calculated using LOWESS (red) from basalt sample. The figure on the left is the spectra from the first laser shot and the right from the 101:th laser shot.
4
Experiment 1 - Investigation of rind
contribu-tions to the plasma during crater formation
4.1
Objectives
As discussed in Section 2.4, rinds and other layered structures can contribute to the induced plasma through plasma-crater interactions. The previous studies of these phenomena have however been conducted under normal, Earth atmo-spheric pressure and composition. While similar results can be expected from Martian conditions the full extent and impact is unknown. Plasma-crater in-teractions have been shown the potential for introducing rind material into the plasma even after the rind is no longer irradiated. The purpose of this experi-ment is to investigate the degree of rind contributions to the plasma throughout the crater formation, using Martian atmospheric conditions.
4.2
Execution
In order to simulate a rock sample with a rind a dolomite slab covered with an aluminum foil was used. Dolomite was used due to its simple chemical composition; CaMg(CO3)2. Since no aluminum is present in the dolomite one
can assume that if significant amounts of aluminum is present in the plasma, it must originate from the aluminum foil. As discussed in section 2.5, calcium is resistant against fractionation that may occur during the crater formation. This predicts a more consistent contribution from the calcium even in deep craters. The aluminum foil was attached to the dolomite sample by the use of clamps. The use of a binding agent could allowed for a better interface between the two layers but doing so would introduce a third layer thus complicating the data analysis.
Since the two layers does not share any major elements an inverse correlation was expected between aluminum on the one hand and calcium and magnesium on the other. It was therefore important to choose aluminum lines which are robust with respect to influence from neighboring lines belonging to calcium and magnesium. If the aluminum lines were sensitive to influence from calcium and magnesium the inverse correlation would be weakened. The ideal aluminum line should be strong and have a large separation to any strong calcium and magne-sium lines. The same properties should ideally also apply to the spectral lines used to represent calcium and magnesium. This is however of less importance since the objective is to study the rind contributions rather then determine the composition of the dolomite slab. The inclusion of the calcium and magnesium line will however be useful in comparing the temporal evolution. The spectral lines were selected from the NIST atomic spectra database [19] and can be found in Table 4.2.
The sample was continuously irradiated using 250 shots allowing a crater to form. The first 205 complete spectra was used to study the spectral evolution. This process was repeated a total of five time creating crater from sample sur-faces not previously irradiated. In order to study the composition of the plasma the intenisty of five spectral lines was studied. The spectral lines are listed in Table 4.2.
Table 4.1: Beam configuration used in experiment 1
Parameter Value
Beam diameter (µm) 1400
Pulse energy (mJ) 385
Irradiance (GW/cm2) 5
Table 4.2: Spectral lines used in experiment 1
Element Wavelength Transition Spectrograph channel
Al II 281.61 nm 3s3p - 3s4s 481 Al I 308.82 nm 3s4p - 3s6d 608 Al II 623.17 nm 3s4p - 3s4d 2156 Ca II 317.93 nm 3p64p - 3p64d 651 Ca II 393.37 nm 3p64s - 3p64p 1011 Mg I 383.23 nm 3s3p - 3s3d 962 Mg I 517.27 nm 3s3p - 3s4s 1618
4.3
Results
The evolution of the total spectral intensity is found in Figure 4.1. It is clear from the data that the overall intensity of the spectra decreases significantly throughout the experiment. Lower intensities makes the measurements more sensitive to instrumentation noise and other errors. Larger shot-to-shot devi-ations can be expected during the later part of the experiment. Two mea-surements clearly deviate from the norm, shots 31 an 32 have a lower over all intensity and extremely large shot to shot variations. These two factors must be taken into account in the continued analysis.
Figure 4.1: Evolution of the total, baseline subtracted spectral intensity from exper-iment 1.
4.3.1 Aluminum
The data for the 281.61 nm aluminum line is found in Figure 4.3. Two clearly de-viating measurements can be seen at shot 31 and 32. These measurements have a somewhat lower mean intensity and exhibits an unusually large confidence interval. The 281.61 nm aluminum exhibits a small increase in line intensity during the first 20 shots be for stabilizing. The intensity remains stable up around shot 50 when a slow decrease in intensity begins. From shot 100 up until shot 130 the intensity is relatively constant with the trend dominated by shot-to-shot variations. From shot 130 a slow decrease in line intensity is seen. The spectral data for the 308.82 nm aluminum line is presented in Figure 4.4. As for the 281.61 nm line two clearly deviating measurements can be seen at shot 31 and 32. These measurements have an somewhat lower mean intensity and exhibits an unusually large confidence interval. During the first 20 shots the line intensity increases significantly. From shot 20 to 50 the intensity is essentially constant and the variations seen are shot-to-shot variations. From shot 60 up until shot 100 the intensity drops noticeably. After shot 100 only shot-to-shot variations remains.
The aluminum line at 623.17 nm is found in Figure 4.5. During the first 50 shots the intensity is constant only exhibiting shot-to-shot variations. From shot 50 until shot 100 a significant decrease in line intensity is seen. Between shots 101 and 140 the intensity continues to decrease but at a slower rate. From shot 141 until the conclusion of the experiment the line intensity continues to decrease in a exponential pattern.
Figure 4.2 compares the evolution of the three aluminum lines. The 623.17 nm clearly has the greatest range and the most dynamic behavior but the same large scale evolution can be seen for the 281.61 nm line though exhibiting a much dynamic range smaller range. For 308.82 nm the decrease in intensity seen from shot 50 clearly coincides with the decrease seen for the other line. From shot 100 the 308.82 nm share no large scale features but some of the larger shot-to-shot variations coincides fro all three lines.
Figure 4.2: Comparison of the evolution of the aluminum spectral lines, average intensity, relative to maximum intensity of spectral line.
Figure 4.3: Evolution of Al II spectral line at 281.61 nm, error bars represent the 95 % confidence interval.
Figure 4.4: Evolution of Al II spectral line at 308.82 nm, error bars represent the 95 % confidence interval.
Figure 4.5: Evolution of Al II spectral line at 623.17 nm, error bars represent the 95 % confidence interval.
4.3.2 Calcium
The intensity data of the 317.93 nm calcium line is found in Figure 4.7. During the first 10 shots the line intensity exhibits a small decrease. From shot 11 to shot 50 the line intensity is very low, in some cases zero. From shot 50 up until shot 100 a rapid increase in line intensity as well as an increase in the shot-to-shot variations is seen. From shot 101 and up until the conclusion of the experiment the line intensity continues to increase but a slower rate then previously.
Figure 4.8 shows the evolution of the 393.37 nm calcium line throughout the crater formation. During the first ten shots the line intensity decreases before stabilizing at a intensity of approximately 3000 a.u. The large error bar for the data from shot 32 indicates a high shot-to-shot variation of the measurements. From shot 50 to shot 100 the line intensity increase rapidly, roughly tripling the intensity. After shot 100 a slow gradual increase in intensity is seen.
As can been seen in Figure 4.6 both calcium lines show many similarities both on a small scale and the overall evolution. During the last 50 shots it is clear that many of the larger shot-to-shot variations coincide for the two spectral lines. The 317.93 nm calcium line exhibit the most dynamic behavior.
4.3.3 Magnesium
The data of the 383.23 nm magnesium line is found in Figure 4.10. During the first three shots the line intensity exhibits a small decrease. From shot 4 to shot 50 the line intensity is in the range of 500 a.u. From shot 50 up until shot 100 a rapid increase in line intensity as well as an increase in the shot-to-shot variations is seen. From shot 101 and up until the conclusion of the experiment the line intensity continues to increase but a slower rate then previously.
Figure 4.11 shows the evolution of the 517.27 nm magnesium line throughout the crater formation. During the first two shots the line intensity decreases before stabilizing at a intensity of approximately 300 a.u. The large error bar for the data from shots 31 and 32 indicates a high shot-to-shot variation of the measurements. From shot 50 to shot 100 the line intensity increases rapidly. After shot 100 a slower, gradual increase in intensity is seen.
Figure 4.9 shows a comparison of the two magnesium lines. It is clearly visible that there exists many similarities between the two lines. The over all evolution of the two lines is identical. Many of the shot-to-shot variations seen in the second half of the experiment coincides fr the two lines. What separates the two line is the larger dynamic range of the 383.23 nm line which is accompanied by a larger shot-to-shot variations
Figure 4.6: Comparison of the evolution of the calcium spectral lines, average inten-sity, relative to maximum intensity of spectral line.
Figure 4.7: Evolution of Ca II spectral line at 317.93 nm, error bars represent the 95 % confidence interval.
Figure 4.8: Evolution of Ca II spectral line at 393.37 nm, error bars represent the 95 % confidence interval.
Figure 4.9: Comparison of the evolution of the magnesium spectral lines, average intensity, relative to maximum intensity of spectral line.
Figure 4.10: Evolution of Mg I spectral line at 383.23 nm, error bars represent the 95 % confidence interval.
Figure 4.11: Evolution of Mg I spectral line at 517.27 nm, error bars represent the 95 % confidence interval.
4.4
Conclusions
The comparison found in Figure 4.12 reveals several similarities in the evolution of the three elements. The three stages of decreasing intensity of the 623.17 nm aluminum line coincided with the increase in intensity for the two calcium lines as well as the two magnesium lines. The inverse correlation is consistent with the type of sample that was used in the experiment.
During the initial stage only the aluminum foil is irradiated. Under ideal conditions only aluminum lines should be present at this stage. The 317.93 nm line has several measurements between shots 1 and 50 which exhibit a zero intensity. The two magnesium lines have non-zero intensities but exhibit inten-sities in the range of 500 a.u. which can be explained by residual baseline and contributions from neighboring lines. The 393.37 nm calcium line has a much higher intensity which can not be explained by residual baseline emission but rather by influence from the strong aluminum lines at 394.40 nm and 396.15 nm. This clearly illustrates the importance of choosing robust spectral lines for this type of analysis.
From shot 50 up until 100 the intensity of the aluminum line drop signif-icantly while the calcium and magnesium lines increase in intensity. This is due to a gradual penetration of the aluminum layer. This phenomena could be observed between shots 50 and 110. Since the beam has a Gaussian intensity profile the penetration depth is greater in the central region as compared to the edge of the beam. As more and more of the foil is penetrated more dolomite is ablated at the expense of aluminum.
After the aluminum foil has been completely penetrated and only dolomite is irradiated by the laser, ideally no aluminum should be present in the plasma. This is clearly not the case since none of the aluminum lines exhibit intensities close to zero after the foil has been completely penetrated. For the 281.61 nm and 308.82 nm aluminum lines this is due to interference from neighboring spectral lines. The 623.17 nm has a better separation to other strong lines and is the central peak of a triplet, this increase the robustness of the line significantly. Despite this, a clear peak at 623.17 nm is still present at the conclusion of the experiment, as can be seen in Figure 4.13. This is in agreement with the findings in Ref. [12] which shows that plasma-crater interactions can introduce materials to the plasma that are not present in the irradiated surface. After shot 100 a slow decrease in intensity for aluminum and a corresponding increase in the intensity of calcium and magnesium can be seen. This behavior can be explained by either a decrease in the aluminum content or a more efficient ablation of the dolomite due to crater interaction. The decrease in total spectral intensity which can be seen in Figure 4.1 is contra-indicative of a more efficient ablation occurring. A smaller contribution of aluminum is a more probable explanation. Even if the plasma retains enough heat to ablate the aluminum from the crater wall the portion of the crater wall which contains aluminum will decrease as the crater depth decreases. As the crater depth increases, the plasma temperature at the rim will decrease despite the confinement of the crater. For sufficiently deep craters the temperature at the rim will eventually be too low to effectively ablate aluminum.
The slightly higher intensities of both calcium and magnesium during the first three shots could be caused by surface contamination. The same sample was used to several experiments and dolomite dust or condensate matter could
be present during the first few shots. This is supported by the presence of dust on internal surfaces of the sample chamber and the sample surface after the experiment was concluded. The reason for the increase in line intensity seen during the first 20 shots for the 281.61 nm and 308.82 is unknown. Changes in the ablation efficiency during the initial crater formation could result in similar behaviors.[11] This is however counter-indicated by the fact that the 623.27 nm line show no such increase.
The conclusion of the experiment is that rinds contribute to the plasma through plasma-crater interactions under Martian conditions. This will result in added difficulties of separating between the composition of the rind and the underlying layers. This will impact the accuracy of quantitative LIBS analysis negatively. While over penetration of the rind decreases the aluminum content this method is both more energy and time consuming and results in a lower depth resolution. The over all result is a lower scientific return in proportion to the invested resources. This indicates the need to find ways of reducing the rind contributions.
Figure 4.12: Comparison of the evolution of selected aluminum, calcium and mag-nesium spectra lines, relative to maximum intensity of spectral line.
5
Experiment 2 - Influence of beam diameter on
rind contributions to the plasma
5.1
Objectives
As in experiment 1, rinds can contribute material to the plasma long after the the rind has been completely penetrated. This fact will greatly reduce the accuracy of the LIBS measurements of samples with rinds by introducing a mixture of the different layers to the plasma. For field applications of LIBS where the sample distance will not be constant, optics must be used in order control the beam properties at the sample surface. In addition to compensating for changes in the sample distance such optics could be used to change the beam diameter on the sample surface. The objective of the experiment is to investigate if using two different beam diameters can improve the measurement accuracy. A larger diameter beam is used to remove the rind and create a crater while a smaller beam diameter is used for the spectral acquisition. By reducing the area of the irradiated surface the plasma will be more confined in the radial direction which could reduce the impact of plasma-crater interactions. Additionally, reducing the beam diameter results in a lower likelihood that the beam will interact directly with the crater walls. The expectation is that the dual beam method will reduce the amount of aluminum in the plasma after the rind is completely penetrated.
5.2
Execution
The dolomite/aluminum sample described in Section 4.2 was used during the experiment. Eight craters with a diameter of 1400 µm was created using the laser. The craters consisted of four groups of two made using 50, 100, 150 and 200 successive shots. A summary of the craters can be found in Table 5.1 In the craters, two laser beams with a diameter of 600 µm and 1000 µm was used for spectral acquisition. The pulse energy was kept constant for all beam configurations. The pulse energy was kept constant for two reasons. Firstly, at a constant energy the total intensity of the spectra should remain relatively constant. Keeping the irradiance constant would lead to much less plasma being induced using the smaller beam diameters. This could affect the accuracy of the measurements negatively. Secondly some LIBS configurations will not allow for changing the pulse energy. The different beam configurations used in the experiment is described in Table 5.2. For each combination of crater and beam diameter four spectra was acquired. The limit of four spectra was chosen so that the depth of the crater would not change by more then 10 % during spectral acquisition.
The results from the experiment 1 is used as a reference. The two data sets are compared by calculating and comparing the 95 % confidence interval. For the data from experiment 1 the confidence interval is represented by the two dashed lines and the mean by the solid line. The confidence interval for the data from this experiment is represented by the error bars.
Table 5.1: Craters parameters used in experiment 2
Crater width (µm) Crater depth (shots)
1400 50
1400 100
1400 150
1400 200
Table 5.2: Beam configuration used in experiment 2
Beam diameter (µm) Pulse energy(mJ) Irradiance (GW/cm2)
1400 385 5
1000 385 9.8
600 385 27.2
Table 5.3: Spectral lines used in experiment 2
Element Wavelength (nm) Transition Spectrograph channel
Al II 281.61 3s3p - 3s4s 481 Al I 308.82 3s4p - 3s6d 608 Al II 623.17 3s4p - 3s4d 2156 Ca II 317.93 3p64p - 3p64d 651 Ca II 393.37 3p64s - 3p64p 1011 Mg I 383.23 3s3p - 3s3d 962 Mg I 517.27 3s3p - 3s4s 1618
5.3
Results
5.3.1 Aluminum
Figure 5.1 shows the data of the 281.61 nm aluminum line. The 600 µm beam exhibits an intensity which is significantly (p<0.05) lower then the reference in the 50 shot crater. In the 100, 150 and 200 shot craters the 600 µm beam resulted in a significantly lower (p<0.01) line intensity.Using the 1000 µm beam to acquire spectra from the 50 shot crater resulted in an intensity which was lower than the reference but not significantly lower (p>0.05). In the 100, 150 and 200 shot craters the 281.61 nm line exhibited significantly lower (p<0.05) line intensity using the 1000 µm beam.
For the 308.82 nm aluminum line, found in Figure 5.2, the line intensities from the 50 shot crater using the 600 µm beam was significantly higher (p<0.05) then reference. In the 100, 150 and 200 shot craters the line intensities from the plasma acquired using the 600 µm beam are significantly lower (p<0.01) than in experiment 1. The intensity is higher sing the 1000 µm beam but not significantly (p>0.05) so. The 1000 µm beam results in significantly lower (p<0.05) line intensity in the 100 and 150 shot craters. In the 200 shot crater the line intensity is significantly lower (p<0.01) then the reference.
For the last aluminum line at 623.17 nm found in Figure 5.3, the 50 shot crater shows no significant difference (p>0.05) from the from the reference using the 600 µm. In the all following crater the 600 µm beam yields significantly lower (p<0.01) line intensities. Using the 1000 µm beam the line intensity in the 50 shot crater is not significantly different (p>0.05) from the reference. The 1000 µm beam exhibits intensities that are significantly lower (p<0.01) then the reference in the 100 and shot crater and significantly lower (p<0.05) in the 150 shot crater. In the 200 shot crater the line intensity is lower but not significantly (p>0.05) so.
Figure 5.1: Comparison of the Al II sp ectral line at 281.61 nm acquired using 600 µ m and 1000 µ m b eam diameter with that acq u ired using 1400 µ m b eam diame ter, the dotte d lines represe n t the 95 % confidence in terv al.
Figure 5.2: Comparison of the A l I sp ectral line at 308.82 nm acquired using 600 µ m and 1000 µ m b eam diameter with that acquired u sing 1400 µ m b eam diame ter, the dotte d lines represe n t the 95 % confidence in terv al.
Figure 5.3: Comparison of the Al II sp ectral line at 623.17 nm acquired using 600 µ m and 1000 µ m b eam diameter with that acq u ired using 1400 µ m b eam diame ter, the dotte d lines represe n t the 95 % confidence in terv al.
5.3.2 Calcium
For the calcium line at 317.93 nm shown in Figure 5.4 the first crater consisting of 50 shots exhibits no significant difference (p>0.05) from the reference line intensity using the 600 µm. In the 100, 150 and 200 shot craters the 600 µm beam results in a line intensity which are significantly higher (p<0.01) than the reference. For the 1000 µm beam a similar behavior is seen with no significant difference in line intensity (p>0.05) in the 50 shot crater and is significantly higher (p<0.01) in the 100, 150 and 200 shot craters.
The 393.37 nm calcium line found in Figure 5.5. In the 50 shot crater the 600 µm beam has a lower intensity but not significantly (p>0.05) so. In the 100 shot crater the 600 µm yields a significantly higher line intensity (p<0.01). The line intensity in the 150 shot crater is significantly higher (p<0.05) while the intensity in the 200 shot crater is not significantly higher (p>0.05). The 1000 µm beam exhibits a higher line intensity in the 50 shot crater but not significantly higher (p>0.05). The line intensities from the 100 shot crater is not significantly higher (p>0.05). In the 150 shot crater the intensity is significantly higher (p<0.05). Finally the line intensity from the 200 shot crater is significantly higher (p<0.01).
5.3.3 Magnesium
The data from the magnesium line at 383.23 nm is found in Figure 5.6. In the 50 shot crater the 600 µm beam yields a significantly lower (p<0.05) intensity compared to experiment 1 while the 1000 µm beam yields no significant differ-ence (p>0.05) in line intensity. The 600 µm beam results in significantly higher (p<0.01) line intensity in the 100 150 and 200 shot craters. Using the 1000 µm beam in the 100, 150 and 200 shot craters all exhibit significantly higher (p<0.05) line intensities.
The data from the 517.27 nm magnesium line is found in Figure 5.7. In the 50 shot crater both beams exhibit no significant difference (p>0.05) in line intensity compared to the reference. In the 100, 150 and 200 shot craters the 600 µm beam exhibits a significantly higher (p<0.01) line intensity. The 1000 µm exhibits a significantly higher (p<0.01) line intensity in the 100 shot crater and higher but not significantly higher (p>0.05) line intensity in the 150 and 200 shot craters.
Figure 5.4: Comparison of the Ca II sp ectral line at 317.93 nm acquired u sing 600 µ m and 1000 µ m b eam diameter with that acquired using 1400 µ m b eam diame ter, the dotte d lines represe n t the 95 % confidence in terv al.
Figure 5.5: Comparison of the Ca II sp ectral line at 393.37 nm acquired u sing 600 µ m and 1000 µ m b eam diameter with that acquired using 1400 µ m b eam diame ter, the dotte d lines represe n t the 95 % confidence in terv al.
Figure 5.6: Comparison of the Mg I sp ectral line at 383.23 nm acquired using 600 µ m and 1000 µ m b eam diameter with that acquired using 1400 µ m b eam diame ter, the dotte d lines represe n t the 95 % confidence in terv al.
Figure 5.7: Comparison of the Mg I sp ectral line at 517.27 nm acquired using 600 µ m and 1000 µ m b eam diameter with that acquired using 1400 µ m b eam diame ter, the dotte d lines represe n t the 95 % confidence in terv al.
