A decade ago few people would have had the fortitude to suggest that high-powered ultrafast lasers were "turnkey." Now they are moving toward widespread use in spectroscopy and in industrial settings. This is a brief look at what has changed in the past decade, what is driving the change, and some of the emerging applications that are enabled by the increasing stability and robustness of ultrafast laser sources.

	For years, since its invention in 1982 by Peter Moulton at MIT Lincoln Laboratory, the workhorse of the ultrafast-laser scientific community has been the Ti:sapphire laser. Able to generate tunable light between 600–1100 nm, the Ti:sapphire laser has been used in fundamental laboratory studies involving frequency-comb spectroscopy, the generation of coherent extreme ultraviolet light, filamentation, and ultrafast laser–material interaction-to name just a few areas of potential interest to readers of this magazine. Initial work on the optical frequency comb technique, for which John Hall and Theodor Hänsch shared one half of the 2005 Nobel Prize in Physics, depended heavily on Ti:sapphire lasers for the generation of the comb.

	The usefulness of the Ti:sapphire laser extends into present work on attosecond (1/1000 of a femtosecond) lasers, in which the Ti:sapphire is used as a pump for molecular interactions generating the final pulse. However, for all its utility and history, the Ti:sapphire laser is generally confined to the laboratory. These systems are still large, complex, and relatively expensive. Although newer models require less care and feeding, Ti:sapphire lasers have always been somewhat difficult to keep stable under changing conditions, requiring a trained operator for stable operation over relatively long periods of time. They are sensitive to temperature and vibration, and therefore are not useful for industrial or mobile applications.

	The telecommunications revolution in the late 1990s and early 2000s led to substantial advances in fiber lasers, in which the gain medium (typically ytterbium, erbium, or thulium) is doped into the fiber. These lasers operate around 1-, 1.5-, and 2-µm wavelengths, respectively. Because all of the gain is in the fiber, and various parts of the laser are coupled with small optical components, the entire system can be made compact and light, with the fiber coil typically defining the limiting dimensions. With the absence of bulk optical elements, ultrafast fiber lasers are much more tolerant of harsh environments. They can be pumped by compact, low-power, and relatively inexpensive and stable laser diodes that are themselves rugged. No beam alignment (or misalignment) occurs, and in general they open up a variety of new operating environments for ultrafast lasers.

	These considerations have driven down the cost and widened the scope of ultrafast lasers. However, considerable development has occurred in the last decade to bring fiber lasers to the current state of the art. In particular, work on maintenance of the polarization and dispersion of the light as it circulates in the fiber cavity has been extensive. The typical dispersion in fiber delays short wavelength light more than longer wavelength light, causing stretching of the frequency components of the pulse as it runs through the fiber. The primary methods to control dispersion use fiber with a configured anomalous dispersion profile, in which slowing of light near absorption bands in the glass reverse this phenomena, delaying the longer wavelength light and recompressing the frequencies in a pulse.

	Still other improvements have improved the performance of ultrafast fiber lasers for particular applications. Laser pulse width can be flexibly controlled by, for example, intracavity fiber Bragg gratings or similar filters. Programmable spatial light modulators are able to cast various shapes, such as multiple spots, lines, and characters onto surfaces, allowing tailored ablation or laser-matter interactions. Improved fibers and pulse control have resulted in increasingly impressive beam qualities, with 

 values as good as 1.3. This indication of the effective divergence of the beam after it exits the fiber means that these ultrafast pulses can propagate in free space for a substantial distance, and focusing qualities will be quite good.

	For purposes of comparison, Table I illustrates typical specifications for laboratory Ti:sapphire lasers and ultrafast fiber lasers.


	Table I: Typical parameters for Ti:sapphire and ultrafast fiber lasers

What Applications Are Enabled by Ultrafast Lasers?

	Ultrafast lasers have a wide array of applications, and as the technology and features improve, the list keeps increasing. Here, we cover a few of the more interesting applications, and some of the higher-volume applications in particular. Each of these topics likely deserves a column of its own, but the treatment here is necessarily abbreviated.

	Laser micromachining and material processing is certainly one of the highest-volume uses of lasers. The benefit of using an ultrafast laser for micromachining is that femtosecond pulses interact in a fundamentally different way with material than do their slower nanosecond, and to a great extent, picosecond cousins. The two physical effects of greatest importance are interrelated (1).

	The first key difference between material interaction with ultrafast and slower lasers is that femtosecond pulses are much faster than the lattice vibration time (10

 Hz) in solids. Because the entire laser pulse occurs much faster than the molecular motion in the structure, no phonons are excited by the laser pulse directly. As a consequence, there is little or no heating of the sample during the laser pulse. Electrons in the material are the only things moving at the timescale of the laser pulse, and the high intensity of the femtosecond pulses causes multiphoton ionization and a cascade of events leading to rapid material removal without substantial heating. In contrast, during ablation with nanosecond pulses, there is substantial material heating. Melting and other physical processes associated with nanosecond pulses yield craters that are much less controlled, and much larger in damage area for the same spot size, compared with ultrafast-laser craters, which typically look as if the material was cleanly removed by a punch, with smooth sides.

	Second, the extremely high fluence of the femtosecond laser pulse allows processing of essentially any material type, including transparent materials. With slower pulses, the photon absorption into a transparent material is typically inversely proportional to the wavelength of light. This advantage led to the early adoption of excimer lasers, with nanosecond pulses and wavelengths at 193 nm and 248 nm, for ablation of materials such as optical glass. These applications persist, but increasingly femtosecond lasers are supplanting excimers as the former become more and more reliable. Because of the mechanism of multiphoton ionization as the initiation to the ablation event, ultrafast lasers in the 1- and 1.5-µm wavelength ranges are absorbed well, even by transparent materials such as glasses.

	As an example, Figure 1 illustrates a multichannel helical mixer with three inlets and two outlets from reference 2. This mixer was made in fused silica using an 800-nm laser with a 50-fs pulse. Fused silica is one of the more difficult ablation targets for nanosecond-class lasers. A programmable three-axis stage was used to guide the laser focus through the fused silica. The helical microchannel has 12 turns in a length of 1200 µm, with a diameter of 150 µm. The cross-section of the flow is 35 ± 1 µm; the authors say that the error in the dimensions of any of the structures can be controlled to 1–2 µm. Following laser ablation, a multistage etch in hydrofluoric acid produces the structure, which is encapsulated in a polydimethylsiloxane film so that it can be connected to the rest of the process.


	Figure 1: A multichannel mixer produced by femtosecond laser ablation in fused silica, observed under an optical microscope. Adapted from reference 2. Reprinted under Creative Commons 4.0 Open Access license.

	Terahertz radiation (also known as submillimeter radiation) is able to pass through a number of opaque materials, particularly nonpolar, nonconductive materials such as textiles, paper, and many plastics. In addition, terahertz radiation, defined as 0.3 to 3 THz, or sometimes more broadly as 0.1 to 10 THz, is non-ionizing, so it is generally safe for humans. Terahertz radiation corresponds to the energy of low-lying ro-vibrational transitions and molecular deformations. Terahertz spectroscopy has uses in security (detection of explosives), medical imaging (3), and numerous other applications. Also, terahertz radiation is highly absorbed by water, making the method sensitive to the density of living tissue. The molecule-specific fingerprint of many complex molecules is unique, and thus there is a high potential for terahertz spectroscopy to find differentiated applications. Compared to microwaves, terahertz waves are much shorter, providing better spatial resolution given the diffraction limit. Unfortunately, terahertz radiation is not as easy to generate as microwaves are.

	One of the issues for terahertz spectroscopy is that there are few sources that generate light in this range, particularly coherent light. Commercial quantum cascade lasers can provide continuous tunable power at the 1–10 mW level, which is useful for absorption spectroscopy. Higher powers, in the 250-mW range, have been achieved in several laboratories. These devices operate at cryogenic temperatures and are somewhat complex and may be difficult to field. Tuning may also be obtained by mixing two near-infrared (NIR) diode lasers that are close in frequency on a semiconductor. This can be used for high-resolution spectroscopy, but the low power output means that scanning through a large spectral range may take minutes or more.

	Ultrafast lasers provide a convenient pulsed source for coherent terahertz radiation. There are multiple methods for producing terahertz radiation, essentially all of them using the ultrafast pulse (~10–250 fs) to generate a similar duration effect in a solid-state semiconductor or crystal. For example, in a photoconductive emitter, ultrafast light interacts with a biased semiconductor material, causing current to flow briefly (picoseconds) in the semiconductor. The semiconductor emits a pulse of terahertz radiation (the fast Fourier transform [FFT] of a picosecond pulse corresponds to terahertz frequencies). This radiation, typically pulsed at high frequency, can be used for illumination of a sample of interest. Terahertz time-domain spectroscopy uses two lasers (or a single laser with a beamsplitter), mixing the return signal to get amplitude and phase information.

	As an example, Figure 2 illustrates the use of terahertz time-domain spectroscopy in the detection of common explosives, with data from reference 4. Explosives RDX (as a pellet), a plastic mixture of RDX, PETN, and polyethylene (which is similar to Semtex H), and a plastic mixture of RDX and polyethylene (which is similar to C4) were each covered with a thin polyethylene sheet and a piece of paper. Additionally, a sample of RDX powder was covered with the polyethylene sheet. Terahertz reflectance spectra were collected and processed using the method detailed in the paper. Obviously strong signals are collected from each of the explosives, and all of the explosives could be differentiated from each other.


	Figure 2: Reflected terahertz signal from common explosives covered with a polyethylene sheet or a piece of paper. Adapted from reference 4. Reprinted under Creative Commons 4.0 Open Access license.

	In the decades since laser-assisted in situ keratomileusis (LASIK) surgery was popularized to reshape the cornea and correct vision, excimer lasers have been the mainstay of the practice. One of the larger concerns that patients have with this procedure has typically been the use of a microblade called a microkeratome to cut the covering of the cornea, exposing the corneal tissue for resurfacing using the excimer laser. Recently, femtosecond lasers have been used to precisely create the cut and avoid use of the microkeratome in the procedure. This so-called "bladeless" LASIK has been in common practice for more than a decade (5).

	Following this success, femtosecond lasers are now routinely used in cataract surgery. In such surgery, a similar incision is made in the eye tissue with the femtosecond laser, and then the laser is used to fragment the clouded cornea, which is removed and replaced with a corneal replacement lens. The laser fragmentation is an improvement compared to the older method of ultrasonic fragmentation, which can lead to energy buildup and heating. The laser method both reduces heating of the eye and softens the lens, making it easier to ultimately remove (6).

	Ultrafast lasers have matured in the past decade, to the point where thousands are in the field performing laser micromachining, security scans, and eye surgeries, among many prominent applications. The development of the ultrafast fiber laser has brought robustness, and years of refinement have brought features such as controllable power, beam shape, and focus. All of these have allowed femtosecond lasers to move out of the laboratory and into the world of applications.

	The author would like to thank Neal Montgomery from Amplitude Systemes for his helpful discussions.

https://doi.org/10.1016/j.tibtech.2016.04.008

(3), (2013). http://doi:10.1117/1.OE.53.3.031202.

 is the CEO of Flash Photonics, an Affiliate Associate Professor at the University of Washington, and a consultant to the spectroscopy industry. Direct correspondence about this column to: 

Articles

Columns: Lasers and Optics Interface

A look at how high-powered ultrafast lasers are moving toward widespread use in spectroscopy and in industrial settings and the developments that are driving these changes


Ultrafast Lasers in Spectroscopy—Moving into the Mainstream

Coherent two-dimensional infrared spectroscopy (2D IR) uses a series of IR femtosecond laser pulses to pump and then probe the response of a system, making it possible to learn much more about the structure and dynamics of molecules than can be seen with one-dimensional IR spectroscopy. The technique's inventor, Martin T. Zanni of the University of Wisconsin-Madison, discussed 2D IR in a 2013 interview (1). Since 2013, Zanni has applied 2D IR spectroscopy to new systems and has started a company, PhaseTech Spectroscopy, Inc., to commercialize the technique.

Can you please give us an update on your research since 2013 (1)?

 provided an overview of coherent two-dimensional infrared (2D IR) spectroscopy and how the mid-IR pulse shaping technology we had developed made it more robust, and discussed some of our work using 2D IR on type 2 diabetes and some simple solar cells. We have covered a lot of new ground since then. On the diabetes project, we discovered a cluster of proteins that we think may be responsible for killing the beta cells in the pancreases of people who have type 2 diabetes, making them more dependent on insulin injections. We started a new line of research studying cataracts, which is some of the first tissue work using 2D IR spectroscopy. In addition, we built the first wide-field 2D IR microscope so that we can now start imaging using the hyper spectral information content of 2D IR spectra. The materials side of my group has now really expanded. For materials, we have transferred some of our technologies from the mid-IR to the visible, and generated a new variant of 2D visible spectroscopy that we call 2D white-light spectroscopy. Perhaps our most impactful accomplishment since 2013 was to start PhaseTech. We have now sold 2D IR and 2D visible spectrometers across the globe.

How did 2D IR help you make your diabetes discovery?

	Almost all people who contract type 2 diabetes have amyloid fibers in their pancreas, where insulin is created. Mysteriously, if amylin fibers are mixed with pancreas beta-cells in a cell culture, the cells are perfectly happy and go on making insulin. If instead, the amylin protein is mixed with cells after the protein has had a chance to age, but not aged quite enough to make fibers, then it is extremely toxic to beta-cells. As a result, lots and lots of people are trying to figure out how the proteins preassemble into something so toxic. We used 2D IR spectroscopy to watch the self-assembly of amylin proteins into amyloid fibers. In a test tube, we used our mid-IR pulse-shaping version of 2D IR spectroscopy to watch the fibers form and discovered that before they aggregate into fibers, they first self-assemble into a cluster of proteins in which the center segment of their sequence lines up into parallel beta-sheets (2). It was fascinating to see this happen because it is very well known that the center sequence, which is formed by the amino acids FGAIL, is critical to diabetes. Humans get diabetes and so do other mammals, such as cats, but lots of other mammals do not, such as mice, rats, and bears. The FGAIL sequence is slightly different from mammal to mammal. Our results suggest that in some species, differences in this sequence prevent this cluster of proteins from forming, thereby preventing the toxic species. There are many more experiments that we must do to further test this hypothesis, but it is an intriguing and potentially important realization.

Was it difficult to apply to coherent 2D IR spectroscopy to tissues in the cataracts work?

	It was surprisingly easy! We had anticipated that the tissues would scatter the laser light horribly, producing a large background, but instead the spectra turned out to be very straightforward to measure. It has been fascinating. In our first published paper (3), we irradiated lenses dissected from the eyeballs of pigs with ultraviolet (UV) light and measured the 2D IR spectra. It turns out that we saw the beta-sheet structures typical of amyloid fibers! We looked very hard with different imaging techniques, but could not see fibers. About 40% of the proteins in the lens are chaperone proteins, suggesting that the amyloid beta-sheet structures are sequestered before they become large enough to see in images. It is fascinating. Work is underway now on human lens tissues.

What was involved in applying the concept of 2D spectroscopy to the visible region of the spectrum?

	The biggest hurdle for us in switching to the visible was conceptual. There are lots of groups utilizing 2D visible spectroscopies: 2D vis spectroscopy is essentially the same technique as 2D IR spectroscopy, albeit at a different wavelength. However, besides wavelength, there is another big difference, which is that visible linewidths are much, much, broader than IR linewidths. Visible linewidths can span hundreds of nanometers. It is very difficult to generate intense laser pulses of comparable width. If the bandwidth of your laser pulse is not at least as broad, if not broader, than the linewidth itself, then you lose all the advantages of 2D spectroscopy because you can't see the cross peaks or 2D lineshapes. We did not enter the field until we had a new idea, which was to use continuum generation for the pulse sequences. Continuum generation has been used as a probe pulse for decades, but it is very weak, which is my guess why no one had tried to use it as a pump pulse before. We generated an entire 2D pulse sequence with it, and it worked great. White light from continuum generation spans nearly the entire visible and near infrared, and so we could then collect 2D spectra, which we now call 2D white-light spectra, with bandwidth that exceeds even the broadest transitions (4). We are now measuring a variety of semiconducting materials using this new source.

What has it been like commercializing coherent 2D spectroscopies?

	It has been very satisfying. My company has helped expand the field quite significantly over the four years that it has been in operation. It has significantly lowered the hurdle for nonexperts to join the 2D field, and made it much faster for experienced users to incorporate pulse shaping into their 2D experiments. It has also given me more of an engineering perspective when we design a new instrument in my research group, because simple instruments are more robust, which leads to higher productivity. It is an exciting time for 2D spectroscopy, because new laser sources are becoming available that will expand the applicability of the technique and lower the cost.

http://www.spectroscopyonline.com/femtosecond-pulse-shaping-enables-rapid-two-dimensional-infrared-spectroscopy?id=&pageID=1&sk=&date=

	(2) L.E. Buchanan, E.B. Dunkelberger, H.Q. Tran, P.-N. Cheng, C.-C. Chiu, P. Cao, D.P. Raleigh, J.J. de Pablo, J.S. Nowick, and M.T. Zanni, 

	(3) T.O. Zhang, A.M. Alperstein, and M.T. Zanni, 

	(4) R.D. Mehlenbacher, T.J. McDonough, M. Grechko, M.-Y. Wu, M.S. Arnold, and M.T. Zanni, 

This interview has been edited for length and clarity. To read more interviews on spectroscopy, please visit: 

www.spectroscopyonline.com/autolist/20/more

Spectroscopy Spotlight

Martin T. Zanni of the University of Wisconsin-Madison is now applying coherent 2D infrared spectroscopy to research on diabetes and cataracts.


New Developments in 2D IR Advance Medical Research and Materials Analysis

We are in a new world where users must specify their analytical instruments. You mean specify a spectrometer? Welcome to a short discussion on the most critical, but often ignored, part of the lifecycle of a spectrometer: Specifying what you want the instrument to do.

	This installment is a short discussion on the most critical, but often ignored, part of the life cycle of a spectrometer: Specifying what you want the instrument to do. Not the salesperson, instrument agent, or supplier, but you. Alternatively, if you want to share the blame with others, consider it you and your colleagues.

	Honesty time! Let us start this painful discussion by asking, have you ever bought an instrument that did not perform as expected? I'm at the front of the guilty queue and I suspect most of you are standing in the line behind me. Buy in haste and repent at leisure, that's the name of the game.

	What do we do with an unsuitable instrument? If it is an absolute lemon, hide it in a cupboard-preferably the farthest away from the boss's office-and then get rid of it at the earliest opportunity. If it sort of works, blame the failure on the supplier as a Monday morning build.

	However, do you ever stop to consider why you are in this sorry state? Let's consider some of the possible (pathetic) excuses:

		The salesperson said it would do the job-worth every word that was written down.

		The salesperson bought me a good meal-short-term gain for long-term pain.

		The sales brochure said it would work-What! You believed the marketing literature?!

		My friend in the other laboratory, site, or company said it was great-I'm not even going there.

	In many companies, this problem is compounded by the end-of-year slush-fund spend. You know what happens, the boss puts his or her head in the lab and says we've got to spend X amount of dollars by the end of the month or we lose it. Panic stations! Your task, should you accept it, is to identify what you are going to buy, obtain three quotes, write a capital request, walk the document around for approval, raise the purchase order, and have an empty box delivered-all within 3–4 weeks. The typical result is an instrument purchased for an undefined use that was bought just to spend budgeted money and avoid unspecified financial consequences. It will sit in a box for an undefined amount of time until the lab gets around to getting it qualified and validated. Will we ever learn?

	You will have noticed that I have not mentioned any regulations or quality guidelines, I merely described what has happened in many laboratories when spectrometers and other analytical instruments are purchased. However, not wishing to disappoint you, here are the regulations and quality standard requirements to consider.

International Organization for Standardization (ISO) 17025

 (1), which is a quality standard for testing and calibration laboratories. Section 5.5 of 

includes the following requirements for analytical instruments:

	"5.5.2 Equipment and its software used for testing, calibration and sampling shall be capable of achieving the accuracy required and shall comply with specifications relevant to the tests and/or calibrations concerned.

	"Before being placed into service, equipment (including that used for sampling) shall be calibrated or checked to establish that it meets the laboratory's specification requirements and complies with the relevant standard specifications."

	Note the phrase "laboratory's specification requirements"-not the supplier's but the laboratory's specification. This phrase implies that there can be a difference between the supplier's specification and that required by the laboratory. Now as a lapsed ISO 17025 assessor, the first version of this quality standard had a footnote to this section that stated that the laboratory's specification may not be the same as the supplier's. It has now been subsumed into the main body of the standard. The important point is that a laboratory does not have to follow the supplier's specification to the letter-what does the laboratory want an instrument to do?

	Let us move on to the United States good laboratory practice (GLP) and good manufacturing practice (GMP) regulations for equipment and see what they say about specifications. The U.S. GMP regulations for equipment are found in section 211.63 (2) and state, "Equipment used . . . shall be of appropriate design, adequate size, and suitably located to facilitate operations for its intended use and for its cleaning and maintenance."

	The corresponding regulations for GLP appear in section 58.61 (3) and are as follows: "Equipment used . . . shall be of appropriate design and adequate capacity to function according to the protocol and shall be suitably located for operation, inspection, cleaning, and maintenance."

	Both U.S. GMP and GLP require adequate design suitable for intended use or function to the protocol, respectively. Intended use has been interpreted as documenting requirements, otherwise how can you determine what the use will be and verify that it works?

 Chapter <1058> on "Analytical Instrument Qualification" (4) has the following statements:

	"The first activity is the generation of a user requirements specification (URS), which defines the laboratory's particular needs and technical and operational requirements that are to be met.

	The subsequent qualification activities necessary to establish fitness for purpose may be grouped into four phases: design qualification (DQ), installation qualification (IQ), operational qualification (OQ), and performance qualification (PQ)."

	It is clear that writing the URS for an analytical instrument is a totally separate activity than the DQ phase. DQ is an activity that confirms that the instrument or instruments you propose to purchase meet the requirements in your URS. You test the requirements in the OQ phase of work. This process was discussed in my "Focus on Quality" column in September 2017 on the new version of 

 <1058> (5). The requirement for a URS followed by DQ is also found in clauses 3.2 and 3.3 in EU GMP Annex 15 on "Qualification and Validation" (6).

	The problem now comes with a further statement in the new 

 <1058>, which states: "It is expected that DQ requirements will be minimal for commercial, off-the-shelf instruments."

	Note these words well. Requirements will be minimal. Not zero, not virtual, not imaginary, not verbal, but minimal. Minimal equals written down, documented, recorded, and formal (approved).

 <1058> says that a URS for an instrument is expected to be minimal, what about any associated software used to control the instrument and also acquire, process, store, and report data? Minimal? I don't think so. We have to factor in data integrity requirements as well as the software functionality for the intended use of the instrument. This process is not an exercise in minimalism. The problem is that all the regulations, such as GMP, GLP, and Part 11, were nearly all written before data integrity became a major issue and regulators are now relying on guidance documents to define their requirements for the integrity of data. This year EU GMP Annex 11 and Chapter 4, on computerized systems and documentation, respectively, will be updated specifically to expand requirements and controls expected for ensuring data integrity into law and not rely on guidance.

	The configuration of software is important for computerized laboratory systems because the intended use of the software, including data integrity, can only be tested when the application settings have been enabled and documented. For example, segregation of duties, enabling record protection, and turning the audit trail on are all important items to configure. A major problem is that instruments and software can be ordered without involving the quality assurance (QA) data integrity function, whose role should be to see if an application is suitable for use in a regulated environment. Some spectroscopic software applications will not save data automatically and will only do so when a user hits the save button. An optional approach to integrity-pick the spectrum you want to save! Look on the bright side, at least you don't have the worry of deleting all those pesky nonconforming spectra.

	The blame game does not just extend to the users in the laboratory. The supplier has an equally important role in the selection process. The new version of 

 <1058> also includes requirement for suppliers (4): "To aid the user, suppliers are responsible for developing meaningful specifications for the users to compare with their needs and aid selection."

	Hmm, meaningful specifications. To illustrate this point I would like to tell you about a benchtop centrifuge that my colleagues and I were involved in qualifying as part of a larger laboratory project. We had to qualify a benchtop centrifuge that operated at room temperature with a fixed rotation speed and a timer. What instrument could be simpler to qualify? However, the rotor speed was specified as 3500 ± 1 rpm. What!? One of my colleagues asked the supplier how this was measured: It was the pulse train to the stepper motor that was recorded when there was no rotor attached.

	I would remind you to look at the role of the supplier a few lines above, especially where it mentions "meaningful specifications." How many of you centrifuge samples without a rotor? Not many I'm guessing. This simple instrument had a specification that was absolute rubbish, but what about more complex ones? Spectrometers are delicate instruments and are made by instrument companies that are based on engineering. Engineers love specifications, for everything-until the marketing department gets their hands on them. The key issue is that you need to understand how the specification was measured and not take it at face value.

	Here's where we get to the Clint Eastwood approach to risk management: Are you feeling lucky? User requirements for commercially available instruments are expected to be minimal. But requirements must also be testable in the OQ using calibrated instruments and traceable reference standards according to the updated version of 

 <1058> (4). As mentioned above, minimal does not mean nothing. We also need our instrument specifications to be scientifically sound and by coincidence so does the U.S. Food and Drug Administration (FDA)! If you look at 21 

 211.160(b) in the GMP regulations you'll also see the phrase "scientifically sound" applied to anything we do in the laboratory (2). Scary stuff, eh?

	Let's assume that we want to purchase an ultraviolet–visible (UV–vis) spectrophotometer. There are many available that come in a range of specifications and budgets. Looking at a high-end instrument, we could have one that can operate over a wavelength range between 175 nm and 900 nm-wow! Imagine the glory of having such a shiny new UV–vis instrument. Hang on a minute. Are you really going to measure down to 175 nm (not forgetting the nitrogen purge)? Step back and ask the following questions: Although the instrument specification is wide, what are you actually going to use the instrument for? What is a realistic wavelength range for your work? Many laboratories may only use a UV–vis instrument over a far narrower range, say between 220 nm and 290 nm.

 17025 discussion earlier in this column, the laboratory specification is 220–290 nm, compared to the instrument specification of 175–900 nm. The former is what you need in your URS, and simply copying the supplier's specification will lead you into scientific and regulatory trouble. Think about what you would have to do about qualifying the wavelength accuracy. Holmium can be used over the range from 240 nm to 640 nm with 14 traceable peaks. But what will you do in the 175–240 nm and 650–900 nm wavelength ranges? The key message is to specify what you will want rather than what the supplier can offer you.

	Associated with this message is the "just in case" argument: For example, you might currently work in the range 240–280 nm, but you might want to widen the range someday. How will this future need impact the specification and how you will verify that in the OQ? My suggestion would be to specify what you need now rather than in the future. When your requirements are updated, look at the new requirements and decide what needs to be done, such as update the URS, extend the wavelength range, and so on.

	This approach also applies to other parts of the instrument specification; you need to specify what you need rather than what is offered, including specifications such as

	This approach has two advantages. The first is to define what you really want and second to avoid purchasing an instrument that you will never use the complete range of functions available.

	If you know how you will use the instrument, writing the specification is much easier. Do you want a research instrument for detailed spectroscopic work or a simpler instrument that just measures absorbance of materials at fixed wavelengths? Thus, a minimal specification for the laboratory is based on what you want the instrument to do and not the supplier specification. From the understanding of intended use will flow the actual specifications based on that usage. While you may use the supplier's specification as a basic starting point for your URS, don't slavishly copy the specification but instead understand how each parameter was measured and adapt accordingly for use. Suppliers have a responsibility here to write meaningful specifications to help all users purchase the right instrument for their needs.

	However, if we continue to avoid writing specifications because "we know what we want" (but don't write it down) then we will still purchase lemons rather than the right instrument for the right job.

	What is the situation if you want to buy the second or third spectrometer of the same make and model? Do you need to write a new specification? Here you should be practical. First, review the current URS for the instrument and ask yourself the following questions:

		Are the requirements still applicable to the current instrument or do they need to be modified?

		Are the current or updated requirements applicable to the proposed instrument? If yes, then order and qualify the new instrument on the basis of the existing URS. If no, then use the current URS as a basis to modify or write additional requirements that reflect the intended use of the new instrument and qualify it on this basis.

	Inertia and fear of the unknown are the factors that stop analytical scientists from writing specification for their instruments. There is also the attitude of management-why waste time, just buy the instrument?

	Although we have discussed quality standards, GxP regulations, and the need to specify your instrument correctly to meet them, the bottom line is less about interpreting clauses in regulations and standards and more about buying the right instrument for the right job. It is investment protection: Resources are scarce and you must spend money wisely. Get the specification right for business reasons and compliance comes as a bonus. You know it makes sense, but does management?

	I would like to thank Paul Smith for his constructive comments during the preparation of this column.

 International Organization for Standardization (ISO) 17025

, "General Requirements for the Competence of Testing and Calibration Laboratories," (International Standard Organisation, Geneva, Switzerland, 2017).

 211, "Current Good Manufacturing Practice for Finished Pharmaceutical Products" (Food and Drug Administration, Silver Spring, Maryland, 2008).

 58, "Good Laboratory Practice Regulations for Non-Clinical Laboratory Studies," (Food and Drug Administration, Silver Spring, Maryland, 1978).

	(4) General Chapter <1058>, "Analytical Instrument Qualification," in 

United States Pharmacopeia 40, 1st Supplement

 (United States Pharmacopeial Convention, Rockville, Maryland, 2017).

	(6) European Commission Health and Consumers Directorate-General, 

EudraLex: The Rules Governing Medicinal Products in the European Union. Volume 4, Good Manufacturing Practice Medicinal Products for Human and Veterinary Use. Annex 15

 is the director of R.D. McDowall Limited and the editor of the "Questions of Quality" column for 

's sister magazine. Direct orrespondence to: 

Focus on Quality

We live in a new world where users must specify what they want their analytical instruments to do. This discussion will help you figure out how to make sure you get what you need.

What! I Have to Specify My Spectrometer?

Currently, there is significant interest in using vibrational spectroscopy techniques for a variety of biomedical applications, and the methods are showing good promise. Karen Faulds, a professor at the University of Strathclyde in Glasgow, has been investigating the application of surface-enhanced Raman spectroscopy (SERS) to the detection of disease pathogens, such as meningitis, and to distinguish related pathogens in a complex matrix. 

You have been doing extensive studies (1–3) on the use of SERS to identify and quantify disease pathogens by detecting their DNA-as an alternative method to using polymerase chain reaction (PCR) with fluorescence spectroscopy. Can you explain briefly how the method works?

	SERS offers significant advantages over other established optical techniques such as fluorescence since it produces a spectrum consisting of a number of sharp bands that are specific to the analyte. The sharp bands discriminate easily from the background, giving high sensitivity and allowing positive identification of an analyte in a mixture. This work involves a new method to detect meningitis pathogens that utilizes the advantages of SERS to detect multiple pathogens simultaneously. Meningitis is usually detected from cerebral spinal fluid and speedy reliable detection of multiple pathogens is vital using the limited sample size available. Currently, bacterial culture is often used and this takes 24 h or more and may not work for all targets. In this research, DNA from the three meningitis pathogens from PCR product was detected using a novel assay that has the potential for faster and more reliable detection-the latter being vital in meningitis assays.

	The assay involves the use of the enzyme lambda exonuclease with the ultimate aim to amplify the signal rather than the amount of target present. A SERS reporter probe was designed that was phosphorylated at the 5' terminus to trigger enzyme recognition, with a target detection region that was complementary to a region of the target sequence and a SERS label. To reduce the possibility of nonspecific adsorption and digestion and allow separation of free probe, a split probe system was designed where the reporter and biotinylated capture probes hybridize to the same target sequence in a sandwich format. The target, reporter, and capture probes were hybridised before being captured on streptavidin-coated magnetic beads. After thorough wash steps to remove excess probe, lambda exonuclease was added and DNA digestion was allowed to occur. Supernatant was separated from magnetic beads and added to citrate-reduced silver nanoparticles followed by addition of spermine hydrochloride and SERS analysis using 532-nm laser excitation. This assay was carried out for the detection of DNA sequences relating to three meningitis pathogens simultaneously. In addition, each of the three pathogens was quantified within the same sample. The quantification of analytes by SERS has long been an issue, and is still very challenging to achieve for even one analyte within a matrix, therefore being able to quantify all three components within a multiplexed, clinical sample is a truly pioneering breakthrough for SERS.

In your method, it was also important that the "reporter" probe not only incorporate a suitable dye label and enzyme substrate, but also that the probe be functionally compatible with the chemistry of the enhancing surface-a colloidal nanoparticle. What is the role of spermine hydrochloride in ensuring that compatibility?

	The silver nanoparticles are made by the citrate reduction of silver nitrate where the citrate acts as both a reducing and capping agent. This results in silver nanoparticles that have a highly negatively surface charge and this is what allows them to remain suspended and isolated in solution, as a result of the repulsive forces between them. To achieve maximum SERS enhancement, the analyte must be as close as possible to the metal surface. However, as DNA is negatively charged it will be repelled by the negative nanoparticles, therefore by adding spermine, a polyamine, it electrostatically adsorbs onto the metal nanoparticles conferring them with a positive charge. This allows the DNA to adsorb onto the metal surface, generating a SERS response. In addition, the spermine also aggregates the nanoparticles, which means any molecules in the interstices between the nanoparticles experience a larger electromagnetic enhancement resulting in increased SERS. Therefore the spermine plays a crucial, dual role.

In a subsequent study, you used this approach to identify three pathogens of bacterial meningitis, in a method that involves labeling each pathogen with a different fluorescent dye (2). Are the spectra from the three dyes easily discriminated?

	Yes, we choose the dye labels for the DNA sequences such that they had distinct spectra with unique peaks so that we could differentiate the presence of all three of the labels, and hence the corresponding DNA sequence, in a mixture, by eye. However, when it comes to quantifying all three, it becomes very challenging to distinguish the three labelled sequences by eye when trying to detect a low concentration of one of the dyes in a background of a high concentration of the other two; to do that, we required chemometrics. In saying that, the fact that we started with three dye signatures that could be visually discriminated made the subsequent quantification easier using the chemometrics.

When testing samples with three pathogens, what is the role of principal component analysis (PCA) in the analysis of the resulting spectra, and in quantifying the pathogens? What is the confidence level that the spectra obtained correspond to the three dyes attached to the different pathogens?

	We chose the dye labels such that they had easily resolved, unique peaks that you could differentiate visually. However, we wanted to be able to quantify each labelled pathogen sequence in the background of the others, that is, to be able to quantify low concentrations of one of the pathogen sequences in the background of high concentrations of the others. This becomes very challenging to do quantitatively and reproducibly by eye. Therefore we collaborated with Professor Roy Goodacre from the University of Manchester, who developed the PCA method that would allow us to quantify each of the three pathogens in a complex spectral matrix of the others. When we carried out the final experiments, the concentrations of the samples were unknown to Kirsten Gracie, who carried out the SERS assay at Strathclyde, and the data generated were analyzed by Elon Correa at Manchester, who also did not know the actual concentrations. Therefore, the assay and the data analysis was effectively carried out using a double-blind approach. The error between the actual and predicted values is very low; the normalized root-mean-square error (NRMSE) associated with the predictions varied between 5.4% and 6.6%, which shows there was excellent correlation between the predicted and actual concentrations.

Why did you choose meningitis for this study? And does meningitis pose any particular challenges to detection?

	We chose meningitis as we had collaborators who were interested in the detection of meningitis and were interested in the ability to detect and quantify pathogens. Meningitis is a dangerous and potentially fatal condition and the specific pathogen needs to be identified rapidly and specifically to allow the correct antibiotic to be administered to rapidly treat the infection and also to reduce the use of broad spectrum antibiotics, given the increasing threat of antimicrobial resistance. Our first study (2) attracted significant media attention and as a result a new collaboration with a clinician interested in using this assay emerged and allowed us to move to using clinical samples to test the assay (3).

	(1) J.A. Dougan, D. MacRae, D. Graham, and K. Faulds, 

	(2) K. Gracie, E. Correa, S. Mabbott, J.A. Dougan, D. Graham, R. Goodacre, and K. Faulds, 

	(3) K. Gracie, D. Lindsay, D. Graham, and K. Faulds, 

This interview has been edited for length and clarity. To read the full interview, visit: 

www.spectroscopyonline.com/detecting-disease-pathogens-surface-enhanced-raman-spectroscopy

Karen Faulds of the University of Strathclyde in Glasgow has been investigating the application of surface-enhanced Raman spectroscopy to identify and quantify disease pathogens by detecting their DNA.


Detecting Disease Pathogens with Surface-Enhanced Raman Spectroscopy

The corrosion resistance characteristics of thermally oxidized stainless steel 316 in an ultrahigh vacuum were examined by exposing this material to an inductively coupled plasma. The CF

–Ar mixed plasma used was generated at 1 mTorr with a constant gas mixing ratio of 9:6:5 sccm. Observations by atomic force microscopy (AFM) demonstrated that after plasma etching oxidized the stainless steel, it retained a smooth surface, while moss-like features formed on the surface of stainless steel that was not thermally oxidized. X-ray photoelectron spectroscopy (XPS) studies revealed that significant amounts of Cr

 remained on the oxidized surfaces after plasma etching, whereas Cr

 appears to be mostly removed from the unoxidized surfaces after plasma etching. In addition, after plasma etching, the concentration of F on the unoxidized sample was greater than that of the oxidized sample in the near surface region according to XPS depth profiling data. Finally, it was found that the 2-nm-thick Cr

 passivation film formed on the oxidized stainless steel surface acted as a barrier to penetration of the fluorine during plasma etching, indicating that the oxidized stainless steel is more corrosion resistant than the unoxidized stainless steel.

	The process of plasma etching is indispensable for the manufacturing of semiconductor devices in the nanoelectronic era (1–5). During the etching process, a mixture of CF

 is commonly used as the reactive gas, and the resultant F ions react with Si

 layers on the surface of the Si wafer, resulting in the creation of a micropattern (6,7). The reactive gas can also react with the process chamber and vacuum components, which are often composed of stainless steel, resulting in corrosion and deposition of metal contamination on the substrate (8–10). Moreover, reactions with radical ions can occur in the pumping system, and the corrosion of the pumping system causes fluctuations of process parameters such as the pumping speed, leading to lower yields and less reproducibility (11,12). Therefore, to obtain high-quality devices, high corrosion resistance levels of the vacuum components during plasma etching with various gases, radicals, and ions are required.

	To improve the corrosion resistance of metals to gases, many researchers have studied various surface treatment technologies by which to passivate metal surfaces (13–19). These approaches involve coating of metal surfaces via physical vapor deposition, chemical vapor deposition, and plasma spraying, among others. In recent work, Wan and colleagues studied Al

-layered CrN coatings on the stainless steel substrate by atomic layer deposition (14). They found that the CrN coating acts as a good blocking layer for diffusion of corrosive substances. However, although these coating methods have demonstrated improvements in corrosion resistance, they require expensive and complicated equipment setups.

	In our previous studies, uniform and smooth Cr

 passivation films were grown on stainless steel surfaces by a simple thermal oxidation method under an ultrahigh vacuum (20–22). The oxide films exhibited remarkable resistance to sorption processes and also served as a diffusion barrier for hydrogen because of their smooth surface and dense structure.

	In the present study, we demonstrate the enhanced corrosion resistance of thermally oxidized stainless steel. Cr

 films were grown on stainless steel surfaces at 500 °C under oxygen partial pressures of 1 × 10

 Torr, after which the surface morphology and chemical composition was investigated after etching with a CF

–Ar plasma. According to atomic force microscopy (AFM) observations, after plasma etching the surfaces of the oxidized stainless steel were much smoother than unoxidized surfaces. X-ray photoelectron spectroscopy (XPS) studies revealed that a large amount of the Cr

 film remained on thermally oxidized surfaces after plasma etching, whereas little Cr

 remained on the surface of unoxidized samples. In addition, according to XPS depth profiling outcomes, the Cr

 films on the passivated stainless steel prevented the formation of metal fluorides and the penetration of F, which is an indicator of corrosion resistance.

	The substrate samples used in this study were commercial-grade 316 stainless steel with compositions of 18.0 wt% Cr, 10.0 wt% Ni, 2.5 wt% Mo, 2.0 wt% Mn and 0.06 wt% C, and balanced Fe. The 0.5-mm-thick stainless steel sheets used here were cut into 1 cm × 1 cm squares and were ultrasonically cleaned in acetone, followed by ethanol. Subsequently, Cr

 passivation films were grown on the stainless steel surfaces in an ultrahigh vacuum chamber with a base pressure of 2 × 10

 Torr by annealing at 500 °C under an oxygen partial pressure of 1 × 10

 Torr. For the corrosion test, the oxidized and the unoxidized stainless steel samples were transferred to a planar inductively coupled plasma (ICP) reactor, as shown schematically in Figure 1. The reactor consisted of a cylindrical chamber with a diameter of 30 cm, a gas controlling system, and a top coil coupled to a 13.56-MHz power supply via a matching network. The samples were etched with CF

–Ar for 10 min with a mixing ratio of 9:6:5 sccm and process pressure of 1 mTorr.


	Figure 1: Schematic diagram of the inductively coupled plasma reactor.

	The surface morphologies of the etched samples were investigated by AFM (Park System XE 100). The chemical states on the sample surfaces and their compositional depth profiles were also analyzed using XPS (Thermo VG Scientific Sigma Probe) combined with Ar

 ion sputtering. The source type for the XPS analysis was Al Kα with a spot size of 400 µm. Binding energies (BEs) were referenced to the C 1s peak from the carbon surface deposit at 284.6 eV.

	To analyze the microscopic structure of the samples, AFM measurements were taken, as shown in Figure 2. Before plasma etching, as shown in Figures 2a and 2b, both the oxidized and the unoxidized samples exhibited smooth surfaces with average roughness (

) of 1.0 and 2.1 nm, respectively, although some small particles from tens to hundreds of nanometers in size on the unoxidized surface were observed. Figure 2c shows that after etching, the oxidized sample surface remained smooth with an 

 of 1.7 nm. In contrast, it can be observed in Figure 2d that plasma etching produced a mossy and roughened surface with an 

 of 4.5 nm for the unoxidized sample, indicating surface erosion.


	Figure 2: AFM images (10 µm × 10 µm) of the stainless steel 316 surfaces before and after plasma etching: (a) the oxidized and (b) the unoxidized stainless steel surfaces before plasma etching, and (c) the oxidized and (d) the unoxidized stainless steel surfaces after plasma etching.

	The chemical composition of the sample surfaces before and after plasma etching was evaluated by XPS analysis. Figure 3 shows the changes in the high resolution spectra of the Cr 2p, Fe 2p, and F 1s photoelectron peaks of the thermally oxidized and unoxidized stainless steel samples before and after plasma etching. In Figures 3a and 3b, the Cr 2p

 spectra can be deconvoluted into three component peaks: those with the highest BEs (CrF

) of 578.2 and 587.6 eV, those at mid-level BEs (Cr

) of 576.7 and 586.0 eV, and those with the lowest BEs (metallic Cr) of 574.0 and 583.0 eV (23–25). As shown in Figure 3a, the Cr 2p spectra for the oxidized samples treated with plasma suggest a large decrease of Cr

, and a small increase of metallic Cr and CrF

 on the surface. In Figure 3b, the peaks from Cr

 on the unoxdized samples after plasma etching are almost not detected. In Figures 3c and 3d, the Fe 2p spectra were deconvoluted into two component peaks: one represents iron oxides such as Fe

, whose BEs are 710.9 and 724.0 eV, and the other denotes Fe–F chemical bonds whose corresponding BEs are 713.9 and 728.0 eV (26–31). A weak Fe 2p

 satellite feature is also observed at 718.8 eV, while no Fe 2p

 satellite feature is observed (27,28). It is clear that before plasma etching, a small amount of iron oxide exists on the surface of the oxidized sample, while a significant amount of iron oxide exists on the unoxidized sample surface. After plasma etching, large increases in iron oxides and fluorides were found on the surface of the unoxidized sample relative to that of the oxidized sample. These results are consistent with previous results from Hara and colleagues that the iron oxides underwent involatile fluorination when exposed to the effluent of the plasma, while the surfaces of the Cr

 films exposed to the plasma were transformed into volatile fluorides or oxyfluorides, which were then pumped away (8). In Figures 3e and 3f, the F 1s spectra exhibit only one peak at 684.8 eV. That peak can be attributed to metallic fluorides such as FeF

, which formed because of the incorporation of F during plasma etching (24,29,31). After plasma etching, the amount of F on the oxidized stainless steel surface was less than that on the unoxidized stainless steel surface, indicating that the oxidized surface is less eroded than the unoxidized surface.


	Figure 3: Comparative XPS spectra of (a-b) Cr 2p, (c-d) Fe 2p, and (e-f) F 1s for the oxidized and unoxidized stainless steel samples before (lower spectrum, black line) and after (upper spectrum, red line) plasma etching, respectively.

	The composition as a function of the depth can also be analyzed by XPS with Ar

 ion beam sputtering. Figure 4 shows the results of the XPS depth profiling for the samples before and after plasma etching. In a comparison of Figures 4a and 4b, after thermal oxidation, the concentration of Cr increases from 18.2 at% to 28.5 at% in the surface region, while that of Fe drastically decreases from 10.3 at% to 2.2 at%, resulting in a 2-nm-thick Cr-rich surface film on the oxidized sample. As shown in Figures 4c and 4d, plasma etching reduces the amount of Cr on the surface by 8.5 at% for the oxidized stainless steel and by 16.0 at% for the unoxidized stainless steel, and increases the amount of Fe by 13.0 and 19.5 at% for each sample.

	Based on our analysis, we can assess that the change of chemical composition on the topmost surface is closely related to the resistance of corrosion. For the unoxidized stainless steel that is vulnerable to corrosion, the amount of Cr is dramatically decreased after plasma etching. Here it is necessary to compare the compositional change of F and O, which are related to product corrosion after plasma etching. The concentration of F for the unoxidized sample is 35.0 at%, which is about three times larger than that for the oxidized case. In contrast, the concentration of O for the unoxidized sample is 20.0 at%, which is about half as low as that for the oxidized sample. These results show that the initial amount of Cr, which mostly exist in the Cr

 before plasma etching, can have a dominant impact on the resistance of corrosion. After plasma etching, the oxidized stainless steel that has enough Cr

 film on the surface can effectively prevent the occurrence of fluoride, demonstrating a lower concentration of F than that of F for the unoxidized stainless steel. We can understand this result because a 2-nm-thick Cr

 film on a stainless steel surface can effectively prevent the incorporation of F during plasma etching (32).


	Figure 4: The XPS depth profiles of (a) the oxidized stainless steel before plasma etching, (b) the unoxidized sample before plasma etching, (c) the oxidized sample after plasma etching, and (d) the unoxidized sample after plasma etching.

	In conclusion, the plasma etching resistance levels of thermally oxidized stainless steel surfaces were analyzed by AFM and XPS with Ar

 cluster ion sputtering. According to AFM observations, the surfaces of the oxidized stainless steel remained smooth after plasma etching, while the unoxidized stainless steel surface became rough and had a mossy appearance, indicating surface erosion. It was found that thermal oxidation of stainless steel increased the amount of Cr on the surface, and decreased the amount of Fe, resulting in the formation of a 2-nm-thick Cr

 film on the oxidized stainless steel surface. In contrast, plasma etching reduced the amount of Cr on the surface and increased that of Fe. As a result, Cr

 on the unoxidized sample surfaces is mostly depleted after plasma etching, while a substantial amount of the Cr

 films remains on the thermally oxidized surfaces. In addition, on the topmost surface, the amount of F in the unoxidized stainless steel was found to be much larger than that in the oxidized stainless steel after plasma etching. These findings demonstrate that stainless steel with a Cr

 passivation film formed by thermal oxidation provides an improvement in resistance to corrosion for conditions similar to that used in the manufacture of semiconductor devices.

	The authors gratefully acknowledge the financial support of the Development Program of Measurement Technology for Advanced Ultrathin Film Processing of the Korea Research Institute of Standards and Science and the R&D Convergence Program of National Research Council of Science and Technology (NST, No. CAP-16-04-KRISS).

Handbook of Advanced Plasma Processing Techniques

 (Springer-Verlag, Berlin, 2000), pp. 257–259.

	(3) M.D. Henry, S. Walavalkar, A. Homyk, and A. Sherer, 

	(4) G. Oehrlein, R. Phaneuf, and D. Graves, 

	(5) H. Kim, J. Woo, Y. Joo, and C. Kim, 

	(6) J. Kim, S. Cho, S. Lee, C. Kim, K. Min, and G. Yeom, 

Although recent steel coating methods have demonstrated improvements in corrosion resistance, they can prove expensive and complicated. In this study, XPS analysis was used to evaluate the corrosion resistance of steel treated with an alternative technique.


Spectroscopy-04-01-2018

Ultrafast Lasers in Spectroscopy—Moving into the Mainstream

New Developments in 2D IR Advance Medical Research and Materials Analysis

What! I Have to Specify My Spectrometer?

Detecting Disease Pathogens with Surface-Enhanced Raman Spectroscopy

X-ray Photoelectron Spectroscopy Analysis of the Enhanced Corrosion Resistance of Thermally Oxidized Stainless Steel

The Role of Mass Spectrometry in Biopharmaceutical Drug Discovery and Development

Market Profile: Process Moisture Instruments

Vol 33 No 4 Spectroscopy April 2018 Regular Issue PDF