Comparative study of two new commercial echellespectrometers equipped with intensified CCD foranalysis of laser-induced breakdown spectroscopy

Mohamad Sabsabi, Vincent Detalle, Mohamed A. Harith, Walid Tawfik, and Hisham Imam

The purpose of this paper is to provide the reader with comparative information about two new com-mercial echelle spectrometers equipped with intensified CCD �ICCD� detectors for laser-induced break-down spectroscopy analysis. We carried out a performance comparison between two commercial ICCD�echelle spectrometers �ESA 3000 �LLA Instruments GmbH, Berlin-Adlershof, Germany� and a Mechelle7500 �Multichannel Instruments, Stockholm, Sweden�� for the determination of the concentrations of Be,Mg, Si, Mn, Fe, and Cu in the same Al alloy samples adopting the same experimental conditions. Theresults show that both systems, despite their differences in terms of resolution, have similar performancein terms of sensitivity and precision of measurements for these elements in an Al alloy matrix at least forthe range of wavelength 280–400 nm studied in this work. © 2003 Optical Society of America

OCIS codes: 120.1880, 140.3440, 300.2140.

1. Introduction

Laser-induced plasma spectroscopy, also known aslaser-induced breakdown spectroscopy �LIBS�, is aform of atomic emission spectroscopy �AES�. LIBSis being used as an analytical method by an increas-ing number of research groups. The growing inter-est in LIBS, due to progressive research andimprovements in the technology of lasers and photo-detectors, particularly in the last decade, has led toan increasing number of publications on its applica-tions, both in the laboratory and in industry.

Undoubtedly, the advent of high-quality solid-statedetectors is revolutionizing the field of atomicspectroscopy.1–3 New optical technologies, whencoupled with new generations of optical detectors,provide powerful tools for plasma diagnostics andspectrochemical analysis. Unlike inductively cou-pled plasma AES or other techniques based on opticalemission spectroscopy �OES�, the LIBS technique re-

M. Sabsabi �mohamad.sabsabi@nrc.ca� and V. Detalle are withthe National Research Council Canada, Industrial Materials In-stitute, 75 de Mortagne Boulevard, Boucherville, Quebec J4B 6Y4,Canada. M. A. Harith, W. Tawfik, and H. Imam are with theNational Institute of Laser Enhanced Science, University of Cairo,Giza, Egypt.

Received 17 January 2003; revised manuscript received 12 May2003.

0003-6935�03�306094-05$15.00�0© 2003 Optical Society of America

6094 APPLIED OPTICS � Vol. 42, No. 30 � 20 October 2003

quires time-resolved detection owing to the transientnature of the plasma. In the initial moments, theplasma emission consists of an intense radiation con-tinuum superimposed with very broadened lines.Thus temporally gating off the earlier part of theplasma is essential for spectrochemical analysis.This dictates an important difference based on time-gated detection of the atomic emission between thedetector requirements for LIBS and other techniquesbased on OES.

Since the early 1990s, echelle-based systems cou-pled with two-dimensional CCD for simultaneousmeasurements of analyte lines at different wave-lengths have been developed and commercialized inthe field of inductively coupled plasma—AES4–9 andused in microwave-induced plasma analysis.10 Inthe past seven years few were used in LIBS,11–19 andit is presently commercialized, to our knowledge, bythree companies.20–22

For analytical spectrochemistry by LIBS, the ap-propriate choice for the experimentalist is based onthe combination of the spectrometer and the detector,which requires a compromise between wavelengthcoverage, spectral resolution, read time, dynamicrange, and detection limit. The combination of spec-trometer and detector is an important factor to con-sider in OES for any plasma characterization oranalytical spectrochemistry experiments. The in-strumental choice of the researcher depends on thetype of measurements to be made. Based on that,

the appropriate system can be designed. For exam-ple, using nonresonance lines of nonmetals such as F,Cl, Br, I, S, N, and O requires working in the near-infrared region ��940 nm�, while the resonance linesof these elements are located in the vacuum ultravi-olet �below 185 nm�. The requirements for an idealspectrometer–detector system to furnish simulta-neous determination of any combination of elementsin the spectrum include:

First, a high resolution of 0.01–0.003 nm to resolvethe lines of interest and avoid interferences.

Second, wide wavelength coverage, typically from165 to 800–950 nm to be able to detect simulta-neously several elements.

Third, a large dynamic range to provide the opti-mum signal-to-noise ratio �SNR� for a large range ofelemental concentrations; the detector needs a widedynamic range, typically 6–7 orders of magnitude.

Fourth, a high sensitivity and a linear response toradiation. The detector has to have high quantumefficiency, particularly in the near-infrared and UVregions, and low noise characteristics. Further-more, for rapid analysis the readout and data-acquisition time should be shorter, at least less thanthe time lap between the laser pulses.

At present, these requirements are in part fulfilledby a combination of echelle spectrometer�intensifiedCCD �ICCD� detector that is on the market. In thiswork we report �for the first time, to the best of ourknowledge� a comparison study between the perfor-mances of two new commercial echelle spectrometersequipped with ICCD detector �LLA InstrumentsGmbH ESA 3000 and Multichannel InstrumentsMechelle 7500� for the quantitative analysis of Alalloys in air at atmospheric pressure in terms of sen-sitivity. Both systems were optimized by the man-ufacturers for the UV-visible range. We chose the Alsamples because of the availability of experimentaldata in our laboratory, making the comparison eas-ier. Furthermore, the poor spectrum of Al does notrequire a spectrometer with high resolution, makingit ideal for this study, since the two systems can beused in our case. The purpose of this paper is to helpthe LIBS users to choose the appropriate system fora selected application in terms of sensitivity. Theperformances related to the resolution or the rate oftransmission will be only mentioned briefly here; de-tails of these parameters are available from the man-ufacturer and can be easily compared by the reader.

2. Experimental Setup

To evaluate the performance of the two-echelle spec-trometer�detector systems �LLA �System A� andMechelle �System B�� for LIBS analysis, we examinedthem in the same experimental conditions. How-ever, LLA was used in the National Research Councillaboratory setup, and the Mechelle was used in theNational Institute of Laser-Enhanced Science exper-iments. Both experimental setups were fully de-scribed elsewhere18,19 and only brief overview will begiven here to make the comparison easier for thereader. Figures 1 and 2 show the experimental

setup used with System A and System B, respec-tively.

The ESA 3000EV from LLA has a focal length of 25cm with a numerical aperture of 1:10. The operat-ing diffraction orders range from 30 to 120. The flatimage plane is 24.85 � 24.5 mm2. This system is acompromise that offers maximum resolution in thewavelength range between 200 and 780 nm �resolu-tion power is above 10,000�. The linear dispersionper pixel ranges from 0.005 nm at 200 to 0.019 nm780. The detector is an ICCD camera, comprised ofa Kodak KAF 1001 CCD array of 1024 � 1024 pixels�24 � 24 �m2� and a microchannel plate type BV2562of 25-mm diameter from Proxitronix coupled with aUV-enhanced photocathode. The Mechelle 7500 hasa focal length of 17 cm with f-number of 5.2. Itprovides a constant spectral resolution of 7500 corre-sponding to 4 pixels FWHM, over a wavelength range200–1000 nm displayable in a single spectrum. A

Fig. 1. Experimental setup at NRC with System A �ESA 3000from LLA�. L, lenses.

Fig. 2. Experimental setup at NILES with System B �Mechelle7500 from Multichannel�. PC, personal computer.

20 October 2003 � Vol. 42, No. 30 � APPLIED OPTICS 6095

gateable ICCD camera, DiCAM-Pro-PCO from Com-puter Optics, with a high-resolution sensor with1280 � 1024 pixels �9 � 9 �m2� coupled to the spec-trometer, was used for the detection of the dispersedlight. The 25-mm microchannel plate is from theDiCAM with a UV-enhanced photocathode. Theoverall linear dispersion of the Mechelle spectrome-ter camera system ranges from 0.0078 �at 200 nm� to0.038 nm�pixel �at 1000 nm�. In both cases a 2-mfused-silica optical fiber �600-�m diameter� mountedon a micro xyz-translation stage is used to collect theemission light from the plasma plume and feed it tothe echelle spectrometer–detector system.

In the experimental setups for System A and Sys-tem B the plasma was produced by focusing 60 mJ ofNd:YAG laser pulses at 1064 nm with 6-ns durationon Al alloy samples. The composition of the samplescan be found in Ref. 18. The laser pulses were suit-ably focused on the Al alloy sample in order to gen-erate plasma of 800-�m spot diameter in both cases.This helps in comparing the performance of bothspectroscopic systems at nearly the same irradiance.It should be noted that there are particles and aero-sols generally present above the sample that arecaused by the ablation of material owing to the laser-sample interaction. Since the threshold of thebreakdown on these particles is lower than that of theair, the likelihood of the plasma generation in the airabove the sample will be increased. Depending onthe size of these plasmas and of the laser beam, thelatter can be partially or completely absorbed bythese plasmas and prevented from reaching the sam-ple. Consequently, both the variation of laser en-ergy density on the sample and the presence of thebreakdown on the particles in the laser path beamwill affect the light emitted by the plasma. Thisincreases the one shot-to-shot variation and de-creases the reproducibility of the measurements. Toavoid these problems, a low repetition rate of 0.2 Hzwas used in both setups. Optical emission from theplasma plume was collected directly by a fibre optic of600-�m diameter positioned close to the plasma�Figs. 1 and 2�, which delivers the light to the en-trance of the echelle spectrometer.

The acquisition of the ICCD of System A was de-layed by 3 �s after the laser pulse was fired. How-ever, to avoid saturation of the detector, theintegration time was limited to 1 �s, and the electrongain was set to a minimum. Similarly for System B,the delay and integration times were 1.5 and 3 �s,respectively. The choice of the delay time and theintegration time is based on the best SNR of the twosystems. Figures 3 and 4 show a spectrum obtainedby Systems A and B, respectively.

3. Results and Discussion

Quantitative spectral analysis generally involves alinear relation from the emission line intensity froman element in the transient plasma from its concen-tration in the target. We investigated a set of eightstandard samples of aluminum alloy to establish cal-ibration curves for five elements �Be, Mg, Si, Mn, and

Cu� by the two systems. We reproduced the mea-surements at five locations on the sample surface inorder to avoid problems linked to sample heterogene-ity. Twenty shots were fired at each location, and

Fig. 3. Single-shot spectrum of aluminum sample obtained bySystem A with zoomed segment showing the beryllium line in theUV region.

Fig. 4. The spectrum is an accumulation of five shots fired on analuminum sample obtained by System B, with the zoomed segmentshowing the beryllium line in the UV region.

6096 APPLIED OPTICS � Vol. 42, No. 30 � 20 October 2003

one measurement is the result of the average for fivelocations. The first 10 shots were used for cleaningthe surface, and the last 10 shots were taken intoaccount for establishing calibration curves to be usedfor the calculation of the limit of detection �LOD� ofeach of the five elements in the Al sample. For thistask we chose the strongest line for each element andbased the calculation on the 3� International Unionof Pure and Applied Chemists definitions. However,we did not use an analytical blank to determine thestandard deviation �, because a region near the linewhere the spectrum is free of emission lines has beenchosen. Using this approach, we determined the an-alytical capability of both systems in terms of theobtained limit of detection. Figure 5 shows the cal-ibration curve of Mn obtained using System A,whereas Fig. 6 depicts the calibration curve of thesame element adopting System B. The Mn contentin the Al samples varies from 0.0025% to 1.09%. Inboth cases the Mn atomic line at 403.1 nm normalizedto the Al line at 305.1 nm has been exploited to buildup the calibration curves. The aluminum line 305.1was chosen because it is not self-absorbed. Asshown in the figures, the two curves are reasonablylinear and comparable in terms of the sensitivity�slope� and the repeatability. In the case of Be, thecalibration for both systems uses the Be �313.0 nm�

normalized to the Al �305.1 nm� line. A comparisonbetween the LOD obtained via the two spectroscopicsystems for the five elements mentioned above, in thesame samples, is given in Table 1. Our results in-dicate that the LOD of these elements obtained bySystem A are similar to those obtained by System B.This is mainly because the two systems were opti-mized for the range of wavelength �250–400 nm�.Outside this range, which is not studied in our con-ditions, when optimization is made for each systemseparately in terms of light collection and SNR somedifferences could be expected.

It should be pointed out that the sensitivity of eachsystem depends on several parameters related to theechelle spectrometer and the detector. For example,the detector is composed of the image intensifier cou-pled to the CCD. The image intensifier comprisesthree main components: a photocathode, a micro-channel plate �MCP�, and a phosphor screen. Thephotocathode converts the incident photons into elec-trons. These electrons are then accelerated towardthe MCP, where multiplication takes place up to anamount dependent on the gain voltage across theMCP. After the MCP stage, the multiplied electronsare accelerated further toward the phosphor screen,where they are converted back into photons ready forthe CCD to detect. Similarly, the input–output fluxof photons through the echelle spectrometer is af-fected by its optical components, namely the disper-sive elements and the mirrors. The output signal Sfor a given wavelength can be written as

S � SphQEpctGVEp QECCD,

where Sph is the average photon flux incident on theICCD photocathode in units photons per second perpixel, QEpc is the average quantum efficiency valueacross the photocathode in units of photoelectrons perphoton, t is the signal integration time, G is the cur-rent gain �amplification of the MCP, typically 1 to3000�, V is the difference between the output andphosphor voltage, Ep is the average phosphor effi-ciency for converting electrons to photons, and QECCDis the average conversion efficiency of phosphor pho-ton signal to CCD electrons.

There is also some dependence on the transmissionefficiency of phosphor light emitted through two op-tical fiber coupler windows.

Fig. 5. Calibration curve of Mn obtained by System A.

Fig. 6. Calibration curve of Mn obtained by System B.

Table 1. Limit of Detection for Various Elements in the SameAluminum Alloys Obtained by Systems A and B

Element

Wavelength ofthe Spectral Line

Used �nm�

LOD

System A LLAEchelle�ICCD

�ppm�

System BMechelle�ICCD

�ppm�

Be 313.0 1.6 0.4Mg 285.21 9 5.5Si 288.16 90 136.6Mn 403.0 65 18Cu 324.75 33 30.4

20 October 2003 � Vol. 42, No. 30 � APPLIED OPTICS 6097

Since the response of these parameters is notequal throughout the spectral range, one systemwill be more sensitive than the other for some re-gions in the spectrum. Furthermore, for an ele-ment that has a rich spectrum �iron, for example�, asystem with high resolution will be more appropri-ate for the application. Another parameter to beconsidered is the dynamic range, which presentssome limits related to the well of the CCD detector.The dynamic range of System A �16 bits� was betterthan that of System B �12 bits�. A high dynamicrange will be preferred in LIBS analysis, particu-larly with echelle spectrometer–ICCD, which pro-vides a large spectrum, including simultaneouslythe strongest lines of the major elements and theweak lines of the trace elements. The dynamicrange is less problematic in the case of CzernyTurner–ICCD configuration than the echellespectrometer–ICCD. This is because the spectralwindow can be adjusted to avoid the simultaneousdetection of the strong lines with weak lines.

4. Conclusion

In summary, we have carried out a comparativestudy between two commercial echelle spectrometersequipped with ICCD UV enhanced detectors in termsof spectrochemical analysis by LIBS of five trace el-ements in Al alloy samples. The two systems areexploited in two typical LIBS setups running underthe nearly the same experimental conditions. Theresults showed that both systems have similar limitof detection of the five investigated elements whenspectral lines in the UV range �280–400 nm� are usedin the analysis.

References1. J. M. Harnly and R. E. Fields, “Solid-state array detectors for

analytical spectrometry,” Appl. Spectrosc. 51, 334A–351A�1997�.

2. F. M. Pennebaker, D. A. Jones, C. A. Gresham, R. W. Williams,R. E. Simon, M. F. Schappert, and M. B. Denton, “Spectroscop-ic instrumentation in the 21st century: excitement at thehorizon,” J. Anal. At. Spectrom. 13, 821–827 �1998�.

3. Q. S. Hanley, C. W. Earle, F. M. Pennebaker, S. P. Madden,and M. B. Denton, “Charge-transfer devices in analytical in-strumentation,” Anal. Chem. 68, 661A–667A �1996�.

4. M. J. Pilon, M. B. Denton, R. G. Schleicher, P. M. Moran, andS. B. Smith, “Evaluation of new array detector for atomic emis-sion spectrometer for inductively coupled plasma atomic emis-sion spectroscopy,” Appl. Spectrosc. 44, 1613–1620 �1990�.

5. T. W. Barnard, M. J. Crockett, J. C. Ivaldi, and P. L. Lundberg,“Design and evaluation of echelle grating optical system forICP-OES,” Anal. Chem. 65, 1225–1230 �1993�.

6. T. W. Hieftje, “The future of plasma spectrochemical instru-mentation,” J. Anal. At. Spectrom. 11, 613–622 �1996�.

7. A. T. Zander, “Continual improvement of instrumentation foranalytical spectrochemistry,” J. Anal. At. Spectrom. 13, 459–461 �1998�.

8. A. T. Zander, R-L. Chien, C. B. Cooper, and P. V. Wilson, “Animage-mapped detector for simultaneous ICP-AES,” Anal.Chem. 71, 3332–3340 �1999�.

9. S. Luan, R. G. Schleicher, M. J. Pilon, F. D. Bulman, and G. N.Coleman, “An echelle polychromator for inductively coupledplasma optical emission spectroscopy with vacuum ultravioletwavelength coverage and charge injection device detection,”Spectrochim. Acta Part B 57, 1143–1157 �2001�.

10. L. Hiddemann, J. Uebbing, A. Ciocan, O. Dessenne, and K.Niemax, “Simultaneous multielement analysis of solid sam-ples by laser ablation-microwave-induced plasma optical emis-sion spectrometry,” Anal. Chim. Acta 283, 152–159 �1994�.

11. H. E. Bauer, F. Leis, and K. Niemax, “Laser induced break-down spectrometry with an echelle spectrometer and intensi-fied charge coupled device detection,” Spectrochim. Acta PartB 53, 1815–1825 �1998�.

12. H. Becker-Ross and S. V. Florek, “Echelle spectrometers andcharge-coupled devices,” Spectrochim. Acta Part B 52, 1367–1375 �1997�.

13. C. Haisch, U. Panne, and R. Niessner, “Combination of anintensified charge coupled device with an echelle spectrographfor analysis of colloidal material by laser-induced plasma spec-troscopy,” Spectrochim. Acta Part B 53, 1657–1667 �1998�.

14. P. Lindblom, “New compact Echelle spectrographs with mul-tichannel time-resolved recording capabilities,” Anal. Chim.Acta 380, 353–361 �1999�.

15. R. G. Scott, S. L. Morgan, R. Hoskins, and A. Oxsher, “Iden-tifying alloys by laser-induced breakdown spectroscopy with atime-resolved high resolution echelle spectrometer,” J. Anal.At. Spectrom. 15, 1133–1138 �2000�.

16. P. Fichet, P. Mauchien, J. F. Wagner, and C. Moulin, “Quan-titative elemental determination in water and oil by laserinduced breakdown spectroscopy,” Anal. Chim. Acta 429, 269–278 �2001�.

17. S. Florek, C. Haisch, M. Okruss, and H. Becker-Ross, “A new,versatile echelle spectrometer relevant to laser inducedplasma applications,” Spectrochim. Acta Part B 56, 1027–1034�2001�.

18. V. Detalle, R. Heon, M. Sabsabi, and L. St-Onge, “An evalua-tion of a commercial echelle spectrometer with intensifiedcharge-coupled device detector for materials analysis by laser-induced plasma spectroscopy,” Spectrochim. Acta Part B 56,1011–1025 �2001�.

19. B. Charfi and M. A. Harith, “Panoramic laser-induced break-down spectrometry of water,” Spectrochim. Acta Part B 57,1141–1153 �2002�.

20. Multichannel Instruments AB, Sweden, http:��www.mul-tichannel.se�.

21. Catalina Scientific Corporation, USA, http:��www.catalinasci.com�.

22. LLA Instruments GmbH, Germany, http:��www.lla.de�.

6098 APPLIED OPTICS � Vol. 42, No. 30 � 20 October 2003