Influence of laser wavelength on LIBS diagnostics applied to the analysis of ancient bronzes
- First Online:
- Cite this article as:
- Fornarini, L., Spizzichino, V., Colao, F. et al. Anal Bioanal Chem (2006) 385: 272. doi:10.1007/s00216-006-0300-1
In this work the influence of laser wavelength upon the analytical results obtained from applying LIBS diagnostics to bronzes was investigated theoretically and experimentally at 1,064 nm and 355 nm. The laser ablation process was modeled for a set of reference samples of quaternary Cu/Sn/Pb/Zn alloys and the difference between plume composition and known target stoichiometry was estimated for both of the wavelengths considered. LIBS measurements were performed on the same set of reference samples and under the same experimental conditions to validate the model at different wavelengths. Results from the application of the model to calculate sample optical properties during laser irradiation, absorption in the plasma and plasma temperature are also presented.
KeywordsLIBS Laser ablation Copper alloys Bronze Modeling
Laser induced breakdown spectroscopy (LIBS) is a well-known, useful tool for performing direct chemical analysis of solid samples. The technique is minimally invasive, which means that it can be applied to the analysis of delicate specimens such as artifacts and artistic objects. Bronze was one of the first alloys developed by ancient metal workers. Their ability to resist corrosion ensured that copper, bronze and brass have remained functional as well as decorative materials used over many millenia to the present day.
Quantitative LIBS analysis of ancient copper-based alloys suffers from problems related to fractionation, which causes marked differences in composition between the ablated plasma and the sample surface. The process of surface evaporation from bronzes is quite complex, mostly due to large differences in the physical properties of the metal constituents (Cu, Sn, Zn, Pb), and this means that the stoichiometry of the plasma analyzed can vary significantly from that of the original surface. When analyzing ancient bronzes, there is also the additional problem of the need to accurately determine the lead content, which is strongly related to the age and intended use of the artwork .
The influence of laser wavelength on fractionation during the process of laser ablation has already been addressed by the analytical community [2, 3, 4, 5, 6, 7, 8]. Ablation mechanisms are known to be influenced by the photon energy of the laser. In general, greater ablation efficiency (amount of mass removed per unit energy), reduced plasma shielding and confinement of sample heating are achieved by using short laser wavelengths (UV) and short laser pulse durations [9, 10, 11, 12]. Fractionation is also known to be a function of the properties of the laser beam (irradiance, pulse duration, wavelength) and the optical properties of the sample; several investigators have reported reduced fractionation when using UV lasers instead of infrared (IR) lasers [4, 13].
Understanding and limiting fractionation mechanisms is of great interest for LIBS quantitative analysis in order to obtain the desired testing accuracy and due to the need to determine the appropriate settings for which the ablated mass vapor is chemically equal to the original sample.
Previously published works on the LIBS analysis of brass alloys have shown that the plume composition matches that of the target more closely if lasers with shorter wavelength are employed and high fluences are used [7, 9, 14]. To our knowledge no data exist on bronze samples. A deeper understanding of the processes involved can also be gained by combining experimental results with models derived from computer simulation.
In this work, the influence of laser wavelength (wavelengths of 355 nm and 1,064 nm were used) upon analytical results of LIBS diagnostics has been investigated both theoretically and experimentally on quaternary bronzes to determine whether improvements are also possible for these alloys. The laser ablation process was modeled for a set of reference samples of quaternary Cu/Sn/Pb/Zn alloys and the difference in plume composition with respect to the known target stoichiometry was estimated for both of the wavelengths considered. The model was based on a one-dimensional heat-flow calculation and was applied to nanosecond laser pulses. The influence of the target properties on the temperature and plume stoichiometry was derived from thermal vaporization processes .
LIBS measurements were performed on the set of reference samples considered for modeling and under the same experimental conditions, so that they could be compared to the model findings and to check whether it is possible to improve the capabilities of LIBS for quantitative analysis.
Standard bronze samples used in this work. Concentrations are given in %wt
Laser ablation of samples was achieved using a Q-switched Nd:YAG laser (model Handy, Quanta System, Solbiate Olona, Italy). Initial experiments were performed with the laser emitting at 1,064 nm, emission at 355 nm was then employed in later tests. In both cases the laser pulse duration was about 8 ns and the repetition rate was 1 Hz. All of the experiments were carried out in air without any control of the surrounding atmosphere. Plasma emission was collected at an angle of about 30° with respect to the laser beam axis. The signal was carried by an optical fiber bundle with a diameter of 0.1 mm. The latter was mounted onto the entrance slit of a Mechelle 5000 spectrograph (Andor, South Windsor, CT, USA). The spectra were recorded using a gated ICCD (iStar DH734, Andor), whose gate aperture was synchronized with the laser burst with an optical trigger. Some preliminary tests were performed to study the temporal behavior of the plasma emission generated. A window suitable for signal maximization was chosen in the temporal range where conditions close to LTE were considered to exist . A delay from the laser pulse of 1,500 ns and an acquisition gate width of 2,000 ns were chosen for laser excitation at 1,064 nm. Data acquisition was performed by accumulating signal over 20 laser pulses at fixed position. Similarly, for a laser operating at 355 nm, the delay from the laser pulse was set to 1,000 ns and the gate width to 1,500 ns. Under these conditions, parameters characterizing the plasma were calculated for both laser wavelengths.
Measurements were carried out at various energy densities over the range 50–250 J/cm2 in order to study the effect of the fluence value on the stoichiometry during the vaporization of metallic targets.
Laser pulse energy was measured behind the focusing lens by means of a Gentec (Markham, Canada) ED-500 energy meter. In order to determine the laser fluence on the sample, the diameters of the laser-produced craters were measured using an optical microscope. Values of 330–400 μm were obtained on average for each series of six measurements.
Atomic lines chosen for detection of the elements under study with their corresponding excitation energies
Peak line emissions, after background subtraction, were determined as average values from a series of six measurements, and the error bars used in the calibration plots were calculated to the 95% confidence interval. The background intensity used for line intensity correction was measured as the average of a wavelength range that was at least 1 nm wide. A range that was free from discrete emission lines and as close as possible to the atomic peak selected for the analysis was chosen when deriving the background relevant to each element.
Theoretical model and results
The interactions between the incoming radiation and the solid sample depend upon numerous variables related to the laser, the sample and the surrounding atmosphere. These variables include wavelength, energy, spatial and temporal profile of the laser beam, and the thermal properties of the sample. The incident beam is partially reflected by the sample surface and partially absorbed by the bulk to a degree that depends on the nature of the target and the temperature it reaches under laser irradiation. The rate of the radiation–solid interaction is also known to depend on the laser wavelength . A theoretical model  has been used to evaluate the difference between the responses of bronze targets to either 355 nm or 1,064 nm radiation.
Maximum temperatures reached in the plasma, obtained from theoretical simulations, calculated for the maximum of the laser pulse (15 ns)
Noting the theoretical results outlined above, we should expect the surface temperature to grow more rapidly for 355 nm radiation. This could affect the relative amounts of the elements in the plasma and minimize (at low fluence) the excess of zinc. As previously observed [18, 27], Zn also evaporates at low surface temperatures, in agreement with its low boiling point and heat of vaporizationcompared to the other elements. Therefore, we expect that the slower the heating process, the higher the zinc content in the plume. At high fluences, the percentage of evaporated Cu, Sn, and Pb increases as higher surface temperatures are reached, where they evaporate more rapidly .
The average background values for the three spectral ranges 275–286 nm, 500–600 nm, 760–860 nm
Selected spectral ranges
Laser excitation wavelength: 355 nm
Laser excitation wavelength: 1,064 nm
The spectra obtained using the IR laser (at 1,064 nm) were acquired using longer acquisition delays (see Experimental section) than those obtained with UV ablation. However, the plasma temperature, as measured using a Boltzmann plot applied to relatively weak Cu transitions, was much higher for IR ablation (13,000±1,200 K) than for UV laser ablation (9,700±800 K). This difference is attributable to the more efficient plasma heating achieved by the IR laser. Plasma electron densities were determined from Stark broadening of the Ca+ ionic line at 393.37 nm, as this element was present as an impurity in some of the samples. The measured electron density for the sample B30 was (3.3±0.7)×1017 cm−3 and (1.0±0.2)×1017 cm−3 for IR and UV laser excitation respectively.
Differences in line intensities were also apparent in these spectra. The spectrum for 1,064 nm irradiation is more intense than the one for 355 nm irradiation in the near-UV region, where ionic lines are preferentially found, and in the near-IR, where emissions from light elements present in air are located. This reflects the higher temperature caused by the greater absorption of the laser light at this UV wavelength, which is rapidly converted into kinetic energy and ionization. On the other hand, the spectrum at 355 nm shows a higher intensity in the visible region. The temperature and electron density reached in this case are most favorable for populating the levels involved in transitions in the visible range.
However, in general, higher atomic copper emission intensity is observed in the three examined ranges for the UV excitation wavelength.
Plume composition versus laser fluence
We investigated the influence of the laser beam irradiance on plume composition from the same bronze samples for different wavelengths.
However, the plume composition of the bronze never reached stoichiometric values at either wavelength, even at high fluences, and it was always too rich in zinc. The fact that the plume is rich in the most volatile element also indicates the thermal process governs the laser ablation in all of the cases examined here [7, 33].
From the literature, the Zn/Cu ratio in plasma produced by ablation of brass materials with 30 ns laser pulses at 248 nm reaches the stoichiometric value for irradiances higher than 0.3 GW/cm2 [7, 14]. For bronze samples, the stoichiometric values for the elements in the plasma are never reached, since the four elements have different thermal properties and so they behave differently with fluence and never attain reciprocal compensation. A nonstoichiometric amount of just one element in the plasma is able to modify the proportions of the other three. A more complex behavior therefore exists due to the higher number of elements involved in the ablation process.
As already seen for 1,064 nm laser irradiation , experimental data normalization using the atomic emission intensity of copper upon UV irradiation did not improve the result.
Internal standardization based on ratios of other elements to Zn has already been shown  to lead to the best calibration for bronze samples for 1,064 nm laser irradiation. The behavior of the Zn concentration in the plume, unlike those of Cu, Sn and Pb, suggests that the presence and quantity of zinc in bronze samples may account for the anomalies in the calibration plots. Standardization on zinc was therefore applied to our experimental data taken at the two different laser wavelengths.
The influence of laser wavelength upon analytical results of LIBS diagnostics of bronzes has been investigated theoretically and experimentally in this work for the laser wavelengths of 1,064 nm and 355 nm, and these results have been compared. The laser wavelengths are those most commonly used in LIBS diagnostics, so the comparison of results between these two cases can help to improve the analytical capabilities of LIBS.
According to model calculations, more limited thermal effects are expected if a laser of a shorter wavelength is used, due to the shorter penetration depth of the light. This is of great significance to the analysis of archeological specimens. The results from the theoretical analysis indicate that the surface temperature grows more rapidly when 355 nm is used rather than 1,064 nm wavelength. This can affect the ratio of the elements in the plasma, which minimizes the excess of zinc in the plume even at low fluence.
The concentrations of the elements in the plasma obtained via calibration-free analysis of the experimental data, show that the zinc content in the plasma is strongly dependent on the laser irradiance for IR excitation. Its concentration decreases with the fluence, approaching the stoichiometric value, while Pb, Sn, and Cu do not exhibit strong dependencies on the fluence. However, lead and tin concentrations remain underestimated. For ultraviolet light, there is almost no dependence of the plasma element's concentration on fluence in the range investigated here. The concentration obtained with the 355nm wavelength is in any case closer to the one of the target with respect to the one obtained with IR light for all considered elements.
It should be pointed out, however, that the plume composition in bronzes for both wavelengths never reaches stoichiometry values even at high fluences, and always remains too rich in zinc content. This is in contrast to what is reported for brass , where stoichiometry is obtained in the plume for a UV laser and at high fluences. A more complex behavior exists in bronze samples due to the higher number of elements involved in the ablation process of the quaternary alloy considered.
The fact the plume is rich in the most volatile element, as theoretically predicted from thermal arguments, allows us to conclude that the thermal process is the dominant effect for the samples and the experimental conditions used here and analyzed by LIBS.
An internal standardization approach based on ratios of the other elements to Zn (previously suggested for IR irradiation ) was then applied to our experimental data taken under UV irradiation, and results were compared with the previous results for IR radiation. In both cases this approach gave the best normalization when calibrating the bronze samples.