LIBS

LIBS Researcher David Hahn wins prestigious Strock Award at FACSS / SCiX

Professor David Hahn of the University of Florida, Photon Machines Advisory Board member, won the prestigious Strock Award at the Federation of Analytical Chemistry and Spectroscopy Societies SCiX meeting this week. The award is given for a selected publication or series of papers of particular significance related to analytical atomic spectroscopy, by the New England Section of the Society for Applied Spectroscopy. Professor Hahn’s work related to particle-plasma interactions in LIBS spectroscopy has been the foundation of much progress in this area, including motivating the work of Photon’s Gregg Lithgow and Steve Buckley. More on the award can be found here (click). Congratulations to David from all of us at Photon Machines!

LIBS 2010 a huge success

Just back from the LIBS 2010 conference last week in Memphis – a great meeting hosted by Jagdish Singh from Mississppi State University.  With over 200 attendees, several short courses (including one by yours truly), 48 podium presentations, and about 100 poster presentations, there was more than enough to see, and there were lots of exciting things to talk about.

One of the clear themes to many of the talks was the increasing importance of chemometrics in most LIBS applications.  At the last International LIBS conference in Berlin, 2008, there were only a few talks that included a substantial portion of chemometrics in the analysis.  Two years later, most papers included some sort of multivariate analysis in their arsenal.  (Here at Photon, we use chemometrics on a daily basis).  This was a good thing to see.  In my short course I discussed both the use and misuse of chemometrics for LIBS.

Another piece of interest to me was the emergence of more applications in industry playing central roles in process control and for key analyses.  As LIBS gains acceptance, we will certainly see more mission critical applications. 

Overall, it was an exciting time for LIBS!  The next US-based meeting will be NASLIBS 2011 in Clearwater, FL, July 18-20, 2011 at the Sheraton Sand Key Resort, co-chaired by yours truly.  The European / Mediterranean Symposium on LIBS (EMSLIBS 2011) will be held in Çe?me-Izmir,  Turkey, hosted by the Izmir Institute of Technology, from September 11-15, 2011.

Check out the program here: http://www.icet.msstate.edu/LIBS2010/LIBS2010_Technical_Program_16July.pdf

Yours in LIBS, Steve

Detectors for LIBS

The subject of detectors for LIBS is quite complicated, encompassing cost, sensitivity, and range. Having used nearly all types of LIBS detectors, unfortunately we can say that there is no magic silver bullet that is useful in every situation. In most cases, there are tradeoffs that need to be made. This short post should suffice in helping you choose a detector for your application.

We will consider simple CCD-based spectrometers, broadband array spectrometers, echelle spectrometers with intensified CCDs, Czerny-Turner spectrometers with intensified CCDs, and PMT-based spectrometers. These have the following attributes:

CCD-based spectrometers: Used in many applications, CCD-based spectrometers are inexpensive and have fairly broad wavelength coverage, depending on the CCD. Typically these cannot be gated faster than a few ms, and because of this, it is difficult to control the data acquisition time or make acquisition very short, which can influence the repeatability of the measurement. Cooled CCDs can have very low dark noise, and thus a very low overall noise floor.

Broadband diode array spectrometers: These spectrometers are arrangements of CCD-based detectors designed to cover a broad wavelength range. Typically uncooled, with multiple-CCD spectra stitched together to form larger broadband spectrum, these systems can provide relatively low cost with multi-element sensitivity. Depending on the configuration, detection limits with these spectrometers can be surprisingly low. At best, these systems generally provide a 0.1 nm resolution.

Echelle spectrometers: Echelles provide broadband coverage which is usually combined with a cooled, intensified CCD that allows gating of spectral acquisition to 10s of nanoseconds. In an echelle, the incoming light is dispersed by a prism and a grating onto the detector, resulting in a 2-dimensional spectral field, the various orders (typically the relatively weaker higher orders) of which are sorted by software to assemble an entire spectrum. Depending on the wavelength region of the emission line, echelle spectrometers with intensified cameras can outperform broadband diode arrays in detection sensitivity by an order of magnitude or more. Resolution is typically 0.05 nm (50 pm) or better. The intensified camera readout noise typically limits the signal-to-noise ratio.

Czerny-Turner spectrometers: These disperse light on a single dimension using a grating, generally using the first order of the grating. On a typical 256 x 1064 intensified CCD array, one thus collects the strongest dispersion of the light (the first order from the grating) and can chose to collect (“bin”) from 1 to 256 rows high of data on the intensified CCD. The strong first order light combined with the choice of binning from 1-256 rows and variation of the gain on the camera intensifier of 1-256 results in an instrument with incredible dynamic range. Unfortunately, limits on the width of the intensified CCD chip are such that gratings with dispersion of about 0.01 – 0.2 nm in a 0.25-meter spectrometer yield only 20-30 nm of spectral range. The grating can typically be rotated on a turret to cover a broad range, but the single-shot spectral range is limited, with the result that simultaneous measurements of multiple elements is typically not possible. With this limitation in mind, the Czerny-Turner with an intensified CCD is one of the most sensitive configurations for LIBS.

Photomultiplier Tube (PMT)-based spectrometers: Photomultipliers are point detectors with much greater potential gain than even an intensified CCD, due to the multiple gain stages – typically they can provide 10e7 signal amplification to the iCCD’s 5 x 10e4 – more than two orders of magnitude improvement. However, as point detectors they need to be implemented either in broad arrays or in with multiple detectors in a Paschen-Runge spectrometer configuration. Hence PMT-based detectors work well once the analyst knows what they are looking for. Another benefit is that PMTs are analog devices that can be continuously read (very fast), providing a dynamic signal that can be particularly useful in understanding events in a laser plasma.

How to sort this out? Obviously there is no clear winner in every category of cost, sensitivity, triggerability, and broadband nature. In these categories the various choices are ranked:

Cost: CCD < Broadband CCD < Paschen-Runge PMT < Czerny-Turner iCCD < Echelle iCCD

Sensitivity: CCD < Broadband CCD < Echelle iCCD < Czerny-Turner iCCD < Paschen-Runge PMT

Triggerability: CCD = Broadband CCD < Echelle iCCD = Czerny-Turner iCCD < Paschen-Runge PMT

Broadband nature: CCD = Czerny-Turner iCCD < Paschen-Runge PMT (depending on # channels) < Broadband CCD = Echelle iCCD

Hopefully this gives the reader a sense for the choices. Please comment here or feel free to contact me if you need help deciding about detectors for a particular application!

What is LIBS?

Laser-Induced Breakdown Spectroscopy (LIBS) is a spectroscopic technique using a laser-generated plasma to ablate and excite a sample, which can initially be in solid, liquid, or gaseous form. Emission generated from the plasma is used to identify material constituents and can be used to identify, sort, and classify materials.

What can LIBS be used for?

LIBS can be used both in-process and in the laboratory for material identification. A very versatile method, it has the primary advantages of

	Rapid analysis
	No sample preparation for most samples
	Sensitive to a wide variety of elements
	Simultaneous reporting of elements

Laser-Induced Breakdown Spectroscopy utilizes a focused pulse from a high-powered laser to create a plasma in or on a solid, liquid, or gaseous media. Some of the energy in the plasma is used to ablate solid or liquid material (if present), and the plasma rapidly expands to form a gas plasma which is used to analyze the ablated particles.

In the core of the plasma, effective temperatures can easily exceed 50,000 K or several eV. During this stage, material in the core of the plasma is vaporized and atomized, and the plasma is typically highly ionized. Depending on conditions, but typically after 0.5 – 1 microsecond, the neutral states of the plasma typically reach local thermodynamic equilibrium. From this time onward, upper electronic states of atoms are thermally populated in Boltzmann equilibrium, such that the emission intensity of the atomic fluorescence I ~ exp(E/kT), where E is the upper state energy level of the fluorescing species.

As the plasma cools, continuum emission from the plasma (Bremstraalung emission, which we see as bright white emission) fades, typically much faster than emission lines from neutral and singly-ionized atomic lines, such that each elemental emission line has a particular optimum in a particular plasma. This optimum depends on the time / temperature history of the plasma, which in turn is dependent on the laser pulse energy and pulse length. For typical 50 – 400 mJ, ~10 ns pulses, the sequence of events in the plasma is shown below; in higher-energy, longer-pulse plasmas, events are shifted to longer times, while for lower energy, shorter-pulse plasmas events are at shifted to shorter times

The time-line of the plasma looks something like:

The emission from the spectra can be then quantified – calibration curves can be obtained by standard peak integration or by use of chemometrics, and/or pattern-matching routines can fingerprint material to determine its type.

Broadband LIBS spectra have a wealth of information:

Calibrations can be used to quantify concentrations, or chemometrics can be used to sort material:

Seminar at Caltech: Discussion of LIBS plasma evolution with time

Last week I had the opportunity to give a LIBS seminar at Caltech to a group of astute engineers and scientists focused on analytical geochemistry. One major vein of questions and discussion that arose was surrounding the transferability of calibrations and the importance of experimental conditions. This is a deep topic indeed and all we can do here is touch on it, but I do believe that there are some easy ways to capture the main issues.

First, realize that LIBS analysis proceeds as the result of two coupled but distinct processes.  The laser ablation process removes material from the sample for introduction to the analytical plasma.  Laser ablation is in and of itself a well-studied process, both by LIBS researchers and those who work on laser ablation ICP mass spec and laser ablation OES.  We know, for example, that the laser may penetrate into some materials, and that in general shorter wavelengths and shorter pulses contribute to better ablation and less fractionation.

The formation of the analytical LIBS plasma, typically in the gas phase above the sample in the ablation plume, occurs as the electron and material density in the plasma exceeds a certain threshold – the plume becomes increasingly optically thick as the developing plasma frequency (determined by the electron density) exceeds the laser frequency.  For this reason, the analytical plasma forms earlier – and ablation stops sooner – with longer wavelength lasers such as 1064 nm YAGs (lower frequencies), compared with shorter 266 nm quadupled YAGs (higher frequencies).

Hence there is a trade-off between ablation on one hand and the analytical plasma on the other, and a balance of energy between the two.  Roughly speaking, if more energy goes into the material ablation, less will be available for the plasma, resulting in a cooler and/or smaller plasma.  The converse is also true.

To wrap up this train of thought, the crux is that varying conditions – in particular the fluence (typically measured in J/cm2) on the sample – change the time-temperature history of the plasma.  The time-temperature history is also influenced by the matrix, laser frequency, etc.

As the plasma cools from incredible temperatures and conditions at the moment of formation, a cascade of events occurs.  Early in time, the plasma emission is dominated by Brehmsstralung emission as decelerating charged particles conserve energy and emit photons.  Highly charged ions recombine and relax, emitting photons.  At relatively early times (0.3 – 2 microseconds in typical plasmas) emission from charged states dominates, while atomic emission is strong from approximately 1 to 10 microseconds (depending upon the atom), and at later times, emission from recombining molecular fragments (CN and C2, for example) is strong.  These numbers are all for atmospheric-pressure LIBS plasmas of moderate energy (50-150 mJ) … your mileage may vary.

An analytical plasma should be one that is in local thermodynamic equilibrium (LTE), which corresponds to conditions in which the electron and gas temperatures are the same, or nearly so, in a local region.  This may be 0.5 to 1 microsecond, commonly.  In a typical plasma, this time to LTE is a function of initial energy and the cooling rate of the plasma – clearly affected by the matrix and by the fluence.  In fact, as an excellent paper has recently argued, the idea of LTE is really a misnomer in a fluidly changing plasma, with steep temperature and diffusion gradients (Christoforetti et al., Spectrochimica Acta B, 65 (2010) 86-95).

At the very least, the considerations thus far lead us to the conclusion that differing initial energies coupled into the analytical plasma lead to a different time-temperature progression – based on the initial size and cooling rate of the plasma, for example.  Hence the optimum detector timing – the delay after the plasma onset to avoid the continuum emission and the gate width when the camera is open – changes with plasma conditions.  For this reason, standard LIBS calibrations are both fluence-dependent and relatively closely matrix-dependent.

Nevertheless, despite these considerations, I would argue (and indeed we and several others have shown) that the Taylor-Sedov blast wave theory does allow useful scaling and similarity solutions for a nondimensional time in laser plasmas.  Changing the energy deposited in the plasma changes the scaling of the plasma decay.  However, knowing the energy in the plasma could allow recovery of the nondimensional time and the implementation of a more universal calibration scheme.

There are myriad ways to approach quantification of LIBS emission, from standard peak integration methods to “calibration-free” methodologies and chemometric methods.  Look for more comments soon on these and other calibration methods right here.

‘Til then, yours in LIBS

Steve

Photon Machines LIBS Blog, Volume 1, issue 1!

What this isn’t: 1) This is not a liberal political blog.  2) This has nothing to do with Nicola Tesla and his Photon Machines.  3) The prefixes Mad- and Womens- have nothing to do with this kind of LIBS.

Welcome to the Laser-Induced Breakdown Spectroscopy (LIBS) blog hosted by Photon Machines.  Here we will delve, wade, and otherwise likely blunder in to all kinds of issues regarding the implementation of LIBS, both for laboratory and industrial uses.  Our biases will be evident and our opinions carry all disclaimers – although they are informed by years of academic and industrial experience.  Hopefully they will be helpful to you, our readers, as you explore the uses of LIBS for analysis of your particular measurement needs.

The benefits of LIBS are many and will be periodically reviewed here.  They include:

• Rapid analysis with minimal sample preparation (no digestion for those of you who practice ICP-MS!)
• Broadband spectra with many simultaneous elemental lines – allowing fingerprinting as well as quantification
• The ability to measure light elements that are difficult or impossible with XRF
• Measurements can be made at a distance (stand-off LIBS)
• Detection limits between 1 and 100 ppm for most elements

All of these mean that there are some pretty interesting applications out there for LIBS.  At the same time, LIBS is NOT for every analysis situation – it does not have the lowest detection limits and the best measurements have an uncertainty between 1 and 4 or 5%.  However, for many situations LIBS has some unique features and new measurement capabilities that add significantly to the analyst’s repertoire.  Not only that, but LIBS is unusually easy, and, we think, fun!

Things that we will talk about in the coming weeks and months include:

• Metallurgical analysis – a nice example of what can be done with LIBS
• Typical approaches to LIBS quantification
• Chemometrics and LIBS
• Pulse energy and laser wavelength considerations
• Spectrometer selection for LIBS

We will do our best to keep the discussion lively and grounded in literature and data.  Please stay tuned, join in often, and stick us on your RSS feeds!  We look forward to an engaging interchange!

Yours in LIBS, Steve Buckley