In September 2013, the United States Pharmacopeial Convention (USP) published a stimulus to the revision process to adopt quality-by-design principles for the development, validation, and operation of analytical procedures used for good manufacturing practice (GMP) analysis. In February 2014, this was followed by a draft Food and Drug Administration (FDA) guidance for industry on validation of analytical procedures. Also, the FDA issued a new draft that updates their 2001 guidance for industry on bioanalytical method validation. Are the approaches the same? What do you think?

I live in London and we have a saying here: You wait a long time for a bus and then three turn up at once. Recently, in method validation a similar thing has occurred and three draft documents for method validation have been issued over a relatively short time span. Two of these method validation documents are intended for laboratories following good manufacturing practice (GMP) principles and one is for bioanalytical laboratories following good laboratory practice (GLP) principles.    

G.P. Martin and colleagues, "Lifecycle Management of Analytical Procedures: Method Development, Procedure Performance Qualification, and Procedure Performance Verification," published in 

				"Analytical Procedures and Methods Validation for Drugs and Biologics," draft Food and Drug Administration (FDA) guidance for industry issued in February 2014 (2); and

"Bioanalytical Method Validation," draft FDA guidance for industry dated September 2013 focused on GLP (3).

We devote this column to reviewing the contents of each of these documents, and comparing and contrasting their approaches to a critical subject for quantitative analysis: the validation of analytical methods or procedures. I will spend most of the column looking at the United States Pharmacopeial Convention (USP) proposal for method development and validation of analytical procedures (1) because this is the most encompassing and radical of the three approaches. Some of these publications refer to analytical procedures and analytical methods — from my perspective the terms are equivalent, and I will use them interchangeably. 

Analytical method development, procedure validation, and transfer to other laboratories to carry out routine analysis are key parts of the traditional analytical process. This starts, if you're lucky, with defining what you want the method to do and ends up with the operational use of the procedure. In the middle is the stuff that should be done properly, but usually is not, as evidenced by how poorly a method operates in routine use. This oversight typically occurs because the method development and validation is performed by one group and the operation of the procedure is carried out by another group. In many instances, the two groups rarely talked to each other except when things went wrong — and blamed each other for the problems. Of course, this never happens in your organization, does it? 

Method validation is not new because there have been a number approaches over the years in both GMP and GLP quality disciplines. Validation of GMP analytical procedures was harmonized with the agreement of the

 International Conference on Harmonisation

 covering the text and methodology of validation of analytical procedures respectively (4,5) in 1994 and 1996, respectively. These two publications were merged into a single document with the revision as 

) in 2005 without changing any text (6). The first bioanalytical methods validation conference was held in 1990 (I was a co-chair of one of the sessions), sponsored by the FDA and the American Association of Pharmaceutical Scientists (AAPS), that drew up the first guidelines for bioequivalence and bioavailability studies typically carried out under GLP (7). This led to the FDA issuing a guidance for industry on bioanalytical methods validation in 2001 (8).   

Validation of an analytical method or procedure, regardless of the GLP or GMP, measures parameters such as precision, accuracy, measurement uncertainty, and limits of quantification. One parameter that I want to focus on is robustness. This is defined in 

The robustness of an analytical procedure is a measure of its capacity to remain unaffected by small, but deliberate variations in method parameters and provides an indication of its reliability during normal usage. 

In the methodology section, there is more information on the topic of robustness (6):			

The evaluation of robustness should be considered during the development phase and depends on the type of procedure under study. It should show the reliability of an analysis with respect to deliberate variations in method parameters. 

If measurements are susceptible to variations in analytical conditions, the analytical conditions should be suitably controlled or a precautionary statement should be included in the procedure. One consequence of the evaluation of robustness should be that a series of system suitability parameters (e.g., resolution test) is established to ensure that the validity of the analytical procedure is maintained whenever used.  

This is the interesting bit — for a document focused on the validation of analytical procedures we have the statement that robustness is actually a job to be done during method development. How many laboratories actually conduct robustness studies? How many laboratories actually conduct robustness studies when under time pressure? However, 

 does say that robustness depends on the method, but then it goes on to mention some typical variables that could be considered for liquid chromatography (LC) procedures such as variations of pH of the aqueous buffer of a mobile phase, variations in mobile-phase composition, or different columns (different lots or suppliers). Therefore, in my view, robustness should be undertaken during the method development phase. There is no discussion in any of the current pharmacopeia general chapters (9–11) or the bioanalytical guidance dealing with validation on method development (8). However, without developing a robust method and knowing how changing a key parameter will impact a method's performance, the work done during method validation could be, and often is, wasted. 

 (6) was to provide guidance in the validation of any analytical procedure, but this document has tended to take on the role of a regulation. Don't think, just do it. 

 has been incorporated into the pharmacopeias; for example, 

 <1225> on validation of analytical procedures (10). Note the chapter numbering: It is an informational (strong guidance) chapter but not a mandatory general chapter (12). 

), there are three general chapters involved with analytical procedures: 					

 <1224> on transfer of analytical procedures (9)

 <1225> on validation of analytical procedures — this is essentially 

 <1226> on verification of compendial procedures (11)

Interestingly, there is no mandatory chapter on the validation of analytical procedures only guidance. The aim of all three chapters is to allow an analytical procedure to become operational in a regulated laboratory as shown in Figure 1. However, there is no consideration of the importance of method development, except where discussed in 

 as mentioned above. As you can see, method development remained outside of the control of pharmacopoeias — despite its importance to the operational method. 

 Figure 1: Current USPâNF general chapters on analytical procedures.			

 the International Conference on Harmonisation has issued three key documents that impact the way that the pharmaceutical industry operates. It changes the industry from reactive to proactive — or should if these documents are implemented correctly. The three documents are 					

 on pharmaceutical quality systems (13): This document describes a quality management system based on ISO 9001, but adapted for the pharmaceutical industry. A key change is the introduction of continual (not continuous as pharma is a regulated industry) improvement. In response to the publication of 

, the European Commission Health and Consumers Directorate-General (EU GMP) chapters 1 (14) and 7 (15) have been issued to incorporate this new approach. 

 on quality risk management (16): This document covers an overall approach to the assessment and management of risk in the pharmaceutical industry. 

 on pharmaceutical development (17): This document is important for the remainder of our discussion because it provides the framework for the approaches that the USP has taken for analytical procedure development, validation, and application. 

 we have to step back in time and outside the pharmaceutical industry. 

A catchy phrase is it not? However, this is key to understanding the overall approach proposed by the USP paper. Therefore, we need to consider the meaning of quality by design (QbD). This concept was outlined first by Joseph Juran, a well-known expert and consultant in quality who stated that "product features and failure rates are largely determined in planning for quality" (18). Put simply, quality must be designed into a product or a process, but quality cannot be tested into it. This is the reason why method development is probably more important than method validation — method validation merely confirms that the development was done correctly.   

Therefore, without adequate quality steps designed into the procedure during the development phase, such as understanding how an analytical procedure works and identifying the major variables, there is not much point in putting it into operation because it would not be very robust or effective. The ineffectiveness would typically be evidenced in the operational use of the method with more out-of-specification (OOS) results compared to a similar procedure in which time and effort has been taken to understand how any analytical procedure method works. This applies in any GMP or GLP laboratory — you reap what you sow. 

Some of the key terms and definitions from 

 are presented in Table I (16). Although 

is focused on the development of a pharmaceutical product, there are several topics that are pertinent to the development, validation, and operation of an analytical method such as a life cycle, design of experiments, and robustness that form the basis of the USP proposed approach. This is why it's important to understand the role and terminology of 

. We will meet the modified versions of these terms for analytical procedures when we look at the new USP approach.  

Table I: ICH Q8 key terms and definitions 			

(1), a USP expert committee published a stimulus to the revision process paper entitled "Lifecycle Management of Analytical Procedures: Method Development, Procedure Performance Qualification, and Procedure Performance Verification." This paper proposes a QbD approach to method development, validation, and operational application of an analytical procedure via a single life-cycle concept. The importance of this stimulus paper is that for the first time there would be a formal link between method development and method validation within a pharmacopeia. Just from the title of the paper we can see the adaptation of two 

 terms: life-cycle management (Q8: life cycle) and procedure performance verification (Q8: continuous process verification). However, there are other terms that are strange, such as procedure performance qualification ­— also, where is method validation in the title?   

In the stimulus to the revision process paper (1), it is proposed that the three existing 

 general chapters dealing with analytical procedure validation, verification, and transfer will be replaced by two general chapters. There will be a mandatory general chapter number <220> dealing with the minimum requirements for method validation and an informational general chapter number <1220> covering the whole life cycle of analytical procedure development, validation, as well as use and modification best practices.  

Overview of the Analytical Procedure Life Cycle

, analytical procedure development, validation, and use and modification is a life-cycle process with defined activities and deliverables. A breakdown of the three-stage process is shown in Figure 2. 

Figure 2: Proposed stages in the analytical procedure life cycle. Adapted from reference 1.			

The first stage of the process is method development and understanding. The last word is important! This goes back to Juran (18): Quality is designed into an analytical procedure during the method development stage and is not tested during the validation part of the life cycle. Therefore, you need to understand how the method works by knowing the key variables and how these influence the analysis: The KISS principle applies here. (KISS? Keep it simple, stupid.) Don't make an analytical method more complex than it needs to be. 

The second stage is procedure performance qualification — eh? This is method validation under another name and has been taken from the FDA's updated "Guidance for Industry" on process validation (19). In essence, this stage demonstrates that the procedure or a modified procedure is fit for the intended purpose of the assay. 

The third stage is the procedure performance verification, which involves checking how the method operates in actual use and that it remains under control. This can be achieved through routine monitoring and charting performance using either Shewart or Cusum plots and reviewing variations in the procedure's performance. If data indicate that the method is not functioning as expected, it should be investigated to see the cause of the variation. Outcomes of an investigation may be a change to the method, which we will discuss later in this column. 

Articles

Focus on Quality

A review of the United States Pharmacopeial Convention (USP) stimulus paper on the revision process to adopt quality-by-design principles for the development, validation, and operation of analytical procedures used for good manufacturing practice (GMP) analysis - and how it compares to a draft guidances by the Food and Drug Administration (FDA) for industry on validation of analytical and bioanalytical procedures.

GLP and GMP Approaches to Method Validation — Going the Same Way?

This article provides a basic overview of the capabilities of transmission gratings optimized for molecular spectroscopy. These gratings offer near 100%, polarization-insensitive efficiency over a broad wavelength range and are ideal for applications such as Raman spectroscopy where the available light intensity from the sample is very low. The design, manufacturing, and performance of the two most commonly used types of transmission gratings — the surface relief transmission grating (SRTG) and the volume phase hologram (VPH) — are described and examples of the two grating types are discussed.

Most diode array–based diffractive spectrometers for the ultraviolet (UV), visible (vis), and near-infrared (NIR) spectral ranges use reflection-type diffraction gratings and are generally optimized for high resolution rather than high optical throughput. During the past 5–10 years, transmission-type gratings have been developed that provide near 100% efficiency for any polarization. These gratings have found widespread use for Raman spectroscopy and astronomy, where very little light is available, but elsewhere the awareness of the distinct benefits provided by transmission gratings seems quite limited. The purpose of this article is to provide a basic overview of the capabilities of transmission gratings optimized for molecular spectroscopy. Specifically, we will describe the two most commonly used types of gratings: the surface relief transmission grating (SRTG) and the volume phase hologram (VPH). 

This section describes the three key performance parameters of a transmission diffraction grating: the angular dispersion, the diffraction efficiency, and the bandwidth of the grating. 

 Figure 1: Transmission grating structure. Key grating parameters: Î = grating period; t = grating thickness; nAVE = average refractive index; and În = refractive index modulation.			

A transmission (or phase) grating consists of a region with a periodic modulation of the refractive index as shown in Figure 1. The period of the index modulation is on the same order of magnitude as the wavelength, and the form of the modulation can be sinusoidal, rectangular, triangular, or even more complex. From the plane wave ray tracing drawn on Figure 1 it can be seen that the plane waves add up in phase for exit angles β that fulfill the following criteria: 

Equation 1 is the general grating equation that is valid for any grating structure. The term 

 is the order of diffraction, and in the remainder of this article we will consider first-order diffraction (

For a certain spectrometer design, the center wavelength as well as the period Λ (or groove density 

 = 1/Λ) for the grating are often given parameters that cannot be freely chosen. Furthermore, a spectrometer is used in a range of wavelengths around the center wavelength and equation 1 determines the angular dispersion of this wavelength range. 

Although the grating equation determines the angular direction (β) of the diffracted light as a function of wavelength (λ), it does not say anything about the diffraction efficiency. The calculation of the diffraction efficiency for a transmission grating geometry generally requires complicated numerical methods such as rigorous coupled wave theory (1) for solving Maxwell's equations. However, Baldry and colleagues (2) approximated analytical equations based on Kogelnik's original work from 1969 (3) to provide a basic relationship between the physical parameters of the gratings and the diffraction efficiency and bandwidth of the grating. This analysis is based on a pure sinusoidal index modulation and is surprisingly accurate for a certain range of gratings. Based on this analysis, it can be shown that a grating's diffraction efficiency, for a fixed grating period and center wavelength, depends on three factors: 					

Figure 2 illustrates the effect of the average refractive index. The figure shows the diffraction efficiency versus grating thickness for three different average indices and constant index modulation. First of all, the diffraction efficiency of both the transverse electric (TE) polarized and the transverse magnetic (TM) polarized wave oscillates between 0% and 100% efficiency with varying grating thickness. However, while the period of the oscillation of the TE diffraction efficiency is almost independent of the average index, the oscillation of the TM diffraction efficiency depends strongly on the average index. For certain values of the average refractive index (for example, 1.285 in Figure 2), the TE and TM can peak at the same grating thickness. This situation is naturally very desirable because it provides a polarization independent efficiency around the center wavelength. 

Figure 2: Diffraction efficiency for TE and TM polarization versus grating thickness for three different average indices in the grating area. Fixed grating parameters: Î»c = 840 nm, Î =  566 nm, and Î± = Î²   (Littrow configuration).			

The effect of the refractive index modulation is illustrated in Figure 3 for the case of a polarization-sensitive design (average index = 1.200) and a polarization-insensitive design (average index = 1.285). From Figure 3 it can be seen that the oscillation frequency of the TE and TM waves increases with increasing index modulation but the relative peak position of TE and TM stays the same. 

Figure 3: Diffraction efficiency versus grating thickness for average indices of 1.2 (left column) and 1.285 (right column) and for three different index modulation strengths. Fixed grating parameters: Î»c = 840 nm, Î =  566 nm, and Î± = Î²  (Littrow configuration).			

The average refractive index controls whether polarization independent ~100% diffraction efficiency can be obtained.

The index modulation controls the ideal thickness of the grating. 

As will be demonstrated in the next section, such nearly 100% efficient, polarization-independent gratings can be produced as both SRTGs and VPHs. The high polarization-independent efficiency is what makes transmission gratings so powerful. 

So far, we have only considered the grating efficiency at a single wavelength. However, a spectrometer needs to analyze a range of wavelengths around the center wavelength. From the article by Baldry and colleagues (2) the full width at half maximum (FWHM) bandwidth of the grating can be calculated by 

It has been assumed that the grating is used in the Littrow configuration where α = β. From equation 2 it can be seen that the bandwidth of the grating is inversely proportional to the grating thickness 

This section provides some insight into the manufacturing methods for SRTG and VPH gratings and presents some of the fine-tuning that can be done to design practical polarization-independent, high-efficiency gratings. 

SRTGs can be manufactured by etching into the surface of a glass substrate using a mask layer to define the periodic pattern. If fused silica is used as the substrate, gratings for wavelengths ranging from UV through NIR can be made. The index modulation goes from ~1.5 in glass to 1.0 in air — that is, an index modulation of ±0.25. The periodic pattern can be created by either a holographic technique or by using a conventional photo-mask. By controlling the pattern exposure and etching process the following parameters can be varied (within certain limits): 					

The shape of the grating "teeth" — both sinusoidal and rectangular is possible

The grating thickness, that is, the etching depth

The duty cycle of the index modulation, that is, the width of the high index portion relative to the total period

The sidewall angle of the grating structures — even asymmetric triangular (blazed) grating structures can be produced

Furthermore, dielectric materials with refractive indices different than the glass substrate can be added to the structure. Figure 4 shows the various parameters that can be used to optimize the grating's diffraction efficiency.  

As discussed in the previous section, a high-efficiency, polarization-independent design requires that the average index, the index modulation, and the grating thickness are controlled and — for the SRTG — this is done by choosing the physical grating parameters appropriately. For instance, for a pure rectangular SiO

 grating profile the average index can be controlled between 1.0 (air) and ~1.5 (SiO

) by choosing the duty cycle appropriately. A 50% duty cycle would yield an average index of ~1.25 and an index modulation of ~0.25. By adding dielectric layers with an index larger than glass, the average index and index modulation can be made even larger than 1.5 and 0.25, respectively. 

When light propagates through the complete SRTG in Figure 4 it passes through three transitions coming from left to right: the air–glass interface; the glass–grating interface; and the grating–air interface. Special care must be taken in the grating design to avoid Fresnel reflections at these interfaces. The first air–glass interface simply requires an antireflection (AR) coating, but the glass-grating and grating-air interfaces must be designed such that the structure itself acts as AR transitions. 

VPHs are typically manufactured in a dichromated gelatin (DCG) that is sandwiched between two glass substrates as shown in Figure 5. The DCG is transparent from ~350 nm through NIR, so VPHs generally cannot be made for UV spectroscopy. The refractive index of DCG can be changed permanently by illumination with laser light. The permanent index modulation is created in the DCG by illuminating the material with an interference pattern (holographic technique). The thickness of the grating is simply the thickness of the DCG layer. The index modulation and average index can be controlled by the exposure intensity or exposure time of the illumination. Unexposed DCG has a refractive index of 1.54 and strongly exposed DCGs yield a minimum average index of ~1.25 and a maximum index modulation of ~0.15. 

Figure 5: VPH grating structural parameters.			

A benefit of the VPH type of gratings is that the grating structure is fully protected between two glass plates, which makes handling and cleaning less of an issue compared to SRTGs.  

Similar to the SRTG, the glass–air interfaces in the VPH must be AR-coated. The VPH process does not provide a way of controlling the glass–grating interfaces. 

With the parameter space available by the manufacturing techniques described above, both SRTGs and VPHs can provide high-efficiency, polarization-independent diffraction gratings. In general, SRTGs provide high index modulation (>0.25) and VPHs provide lower index modulation (<0.15). This means that VPHs typically require thick gratings (

 > 5λ) and SRTGs require thinner gratings (

The bandwidth of the grating is inversely proportional to the thickness of the grating, which means that SRTGs will provide a wider bandwidth than VPHs. 

Table I provides a general overview of the two transmission grating types. 

Table I: Comparison of SRTG and VPH capabilities			

To illustrate how effective transmission gratings can be, we will consider two common spectrometer applications: a Raman spectrometer with a laser wavelength of 785 nm and a UV spectrometer. 

Peer-Reviewed Articles

This article provides a basic overview of the capabilities of transmission gratings optimized for molecular spectroscopy.

Overview of High-Efficiency Transmission Gratings for Molecular Spectroscopy

Chlorophyll and leaf water content are important nutritional indicators of the Gannan navel orange tree. Hyperspectral imaging was used to simultaneously estimate chlorophyll and water content of Gannan navel orange leaves. The genetic algorithm (GA) and successive projections algorithm (SPA) were applied in selecting characteristic wavelengths after average spectra were extracted from the region of interest (ROI). The hyperspectral effective information was extracted using the partial least squares (PLS) model. The results derived from these techniques were then compared. The prediction model of chlorophyll and water content built using SPA on 8 and 14 selected wavelengths (3.15% and 5.15% of the total number of wavelengths) exhibited higher coefficients of determination 

 of 0.98 and 0.93 and root mean square error of prediction (RMSEP) values of 3.40 and 0.04, respectively. The results show that the accuracy of the quantitative analysis conducted by hyperspectral imaging can be improved through appropriate wavelength selection. Although the RMSEP value of SPA-PLS was slightly higher than that of GA-PLS, the SPA-PLS model can be used for applications because it was simpler and easier to interpret.

Chlorophyll and leaf water content indirectly reflect the ability of plant growth and photosynthesis and they are closely related to the nutritional status of the plant, which is one of the most important indicators of the Gannan navel orange photosynthetic capacity and developmental stage. 

 is a green pigment found in cyan bacteria and the chloroplasts of algae and plants. It is an extremely important biomolecule that is critical in photosynthesis, which allows plants to absorb energy from light. Chlorophyll absorbs light mostly in the blue region of the electromagnetic spectrum. However, chlorophyll is a poor absorber of green and near-green regions of the spectrum, hence the green color of tissues containing chlorophyll (1). Traditional measurement of chlorophyll and water in a leaf damages the blade and is time consuming. Typical methods include spectrophotometry (2) and gravimetric methods. 

In recent years, hyperspectral imaging has gained wide acceptance in different fields by virtue of its advantages over other analytical techniques. The most important advantage is its ability to record spectra and images simultaneously. The rapid, nondestructive, and reliable features were first used in military and remote sensing (3), and then were applied in astronomy (4), food (5), agriculture (6), and pharmaceuticals (7). Hyperspectral imaging at visible (vis) and short near infrared (SNIR) regions has been used to estimate chlorophyll or water content (8–10) in leaf samples. 

Among numerous wavelength selection methods, partial least squares (PLS) has been extensively used to enhance the performance of models constructed (11). Genetic algorithms (GA) add complexity to the computation when the number of selected wavelengths increases (12). Sorol and colleagues (13) applied several wavelength selection algorithms to sort informative wavelengths in building a PLS model to predict the degrees Brix (sugar content) of sugar cane juice. They found that the best results were obtained using GA. A novel forward selection method called the successive projections algorithm (SPA) has been used to obtain a simpler model for minimizing wavelength collinearity. Chen and colleagues (14) used SPA to optimize the NIR spectroscopic quantitative model for potable spirits; they obtained a good predictive correlative coefficient of 0.9477 using only 1.17% of the total number of wavelengths. 

Typically, only one variable quantity has been discussed in hyperspectral data of a leaf. In this study, chlorophyll and water content were investigated at the same time for Gannan navel orange leaves. Thus, constructing a robust model using selected informative wavelengths to improve modeling efficiency and prediction accuracy was crucial (15). GA and SPA were chosen as efficient wavelength selection methods for improving the PLS model.  

 Figure 1: Chlorophyll content measurement to select the location 			

Gannan navel orange leaves covering every stage of growth period from 20 navel orange trees were obtained from an orchard in Jian county, Ganzhou city, Jiangxi province, China. The first batch of 58 samples was collected on October 5, 2012, and the second batch of 78 samples was collected on November 1, 2012. Orange chlorophyll content was measured using a SPAD-502Plus chlorophyll meter (Spectrum Technologies, Inc.). The value provided by the meter indicates the relative content of chlorophyll in the leaves. The meter can measure the chlorophyll content of leaves within only a small spot, so five points on the leaves were sampled to measure the content in a given region. The average of the five positions was determined to generate the chlorophyll content as shown in Figure 1.  

The leaf data were collected by the hyperspectral imaging system as shown in Figure 2, which consists of a line-scan spectrograph (ImSpecter, V10E, Spectra Imaging Ltd.), a charge-coupled device (CCD) camera (Hamamatsu, C8484-05G), two halogen lamps (20 W, Osram), and a precision sample stage with a motion controller. The working distance of the hyperspectral lens to the displacement platform was 50 cm, and the light source was 65 cm above the samples. The angles of the two lamps were adjusted to ensure the light source was centered on the stage and to obtain a uniform illumination. A black plastic base plate was placed on the stage. To avoid baseline drift, the hyperspectral imaging system was preheated for 30 min. The line-scan spectrograph collected spectral images over a wavelength range of 382–999 nm with a spectral resolution of 2.8 nm. The sample stage was driven by a stepping motor with a user-defined speed 16 mm/s. The exposure time, camera resolution, and spectral sampling interval were set at 20 ms, 1344 × 1024 and 2.44 nm, respectively. Next, 254 wavelengths were sampled to generate a 1344 × 1024 × 254 hyperspectral image cube. The first batch of 58 leaves and the second batch of 78 leaves were measured separately on October 10, 2012, and November 5, 2012. 

Figure 2: Hyperspectral imaging system.			

) image was obtained by collecting a sample image, a dark image, and a white image, and was processed according to the following equation: 

 is the dark image (with 0% reflectance) recorded by turning off the lighting source with the lens of the camera completely closed, and 

 is the white reference image (PTFE white board with 99% reflectance) (16). The above data collection and correction processes were implemented using software called spectraSENS1.5.  

The 136 samples were randomly divided into a prediction set (45 samples) and a calibration set (91 samples). Table I shows the properties of the samples used for calibration and prediction of the PLS models for both chlorophyll and water content.  

Table I: Descriptive statistics for sample measurement			

A square region of interest (ROI) (17) was defined by a 7 × 7 pixels area that overlaps with the region assessed by the chlorophyll meter. The ROI for water content covers the entire leaf. This is shown in Figure 3. The mean relative reflectance for each image was generated by averaging the spectral responses of each pixel in the respective ROI. In total, 136 mean reflectance spectra of chlorophyll and water were separately obtained from 136 samples.  

Figure 3: Regions of interest of the orange leaf: (a) chlorophyll, (b) 			

GA was applied in the spectral analysis to select several intervals and calculate all PLS model combinations of these selected intervals. The average spectrum of ROI was imported Matlab R2010a and the parameters were set as follows: The number of groups was 30, crossover probability was 0.50, compiled probability was 0.01, and the number of iterations was 100, run 100 times independently. Finally, each output the 0–1 binary coded string, calculated wavelength point identification of the probability of "1."   

Figure 4: NIR average spectra from all 136 samples of the (a) chlorophyll 			

Figure 5 shows the frequency of wavelength selection, 13 optimal wavelengths of chlorophyll were 491, 494, 557, 560, 562, 640, 643, 645, 662, 665, 667, 911, and 913 nm. We chose 63 optimal water content wavelengths: 389, 391, 443, 445, 447, 455, 457, 460, 469, 472, 474, 477, 547, 550, 552, 569, 572, 574, 596, 599, 601, 618, 630, 633, 635, 638, 655, 701, 784, 787, 789, 821, 823, 826, 828, 831, 833, 882, 884, 899, 904, 906, 909, 911, 913, 916, 918, 938, 940, 943, 945, 948, 950, 953, 955, 970, 972, 975, 984, 987, 989, 996, and 999 nm.  

Figure 5: Selected wavelength using GA: (a) chlorophyll, (b) water  			

The SPA is a forward selection method, which begins with one wavelength, and then incorporates a new one at each iteration until a specified number 

 of wavelengths is reached (18). More details on the steps involved in SPA application can be found in previous studies (19–21). The wavelengths selected by the SPA method can be used as the input of the multiple linear regression (MLR), PLS, and other regression algorithms. 

The SPA method was used to select informative individual wavelengths among the spectra (Figure 5). We gained eight optimal wavelengths of chlorophyll: 389, 394, 418, 494, 538, 643, 706, and 884 nm. We obtained 14 optimal wavelengths of water content: 384, 406, 450, 545, 596, 660, 677, 694, 738, 789, 828, 893, 950, and 998 nm.  

Figure 6: Selected wavelength using SPA: (a) chlorophyll, (b) water content. 			

The PLS models were established with the selected wavelengths as shown in Figure 7.  

Figure 7: Predicted values with PLS models: (a) chlorophyll-GA-PLS, (b) water-GA-PLS, (c) chlorophyll-SPA-PLS, (d) water-SPA-PLS.			

Table II shows the number of variables used in each model. The percentage of wavelengths used to develop the chlorophyll-GA-PLS, chlorophyll-SPA-PLS, water-GA-PLS, and water-SPA-PLS models were 5.12%, 3.15%, 24.8%, and 5.51%, respectively. The PLS method is also relatively complex because of its complex algorithm (22). The accuracy of GA-PLS models was higher than SPA-PLS models, but the GA-PLS models used more wavelengths. GA-PLS was considered sufficiently suitable for the prediction of chlorophyll and water content in Gannan navel orange leaves.  

Table II: Results of model predictions based on different algorithms			

Hyperspectral imaging was successfully utilized for the simultaneous detection of chlorophyll and water content in Gannan navel orange leaves. Two wavelength selection methods, GA-PLS and SPA-PLS, were compared. The appropriate selection of wavelengths can enhance the performance of a model. GA-PLS provided better performance than SPA-PLS. GA-PLS was developed with a large number of wavelengths (5.12% and 24.8% of the total number of wavelengths), which makes it relatively more complex compared to SPA-PLS (which only used 3.15% and 5.51% of the total number of wavelengths). However, the GA-PLS approach provided a better prediction accuracy, so it is better suited for industrial application. 

ISPRS J. Photogrammetry and Remote Sensing 

(10) F.M. Danson, M.D. Steven, T.J. Malthus, and J.A. Clark, 

(11)  A.S. Bangalore, R.E. Shaffer, and G.W. Small, 

(12) J. Koljonen, T.E.M. Nordling, and J.T. Alander, 

Here, hyperspectral imaging was used to simultaneously estimate chlorophyll and water content of Gannan navel orange leaves.

Simultaneous Measurement of Chlorophyll and Water Content in Navel Orange Leaves Based on Hyperspectral Imaging

Laser-induced breakdown spectroscopy (LIBS) is an emerging analytical method that has been the focus of substantial research over the last 25 years. The recent emergence of commercially available LIBS systems from major manufacturers is a sign of the maturing of the technology. This column installment is the second of a three-part series focusing on major aspects of LIBS. The first installment set the stage with the basics of the LIBS measurement physics and standard applications. This installment discusses the choices for LIBS hardware in detail, particularly lasers and spectrometers, and illustrates the trade-offs between cost, size, and performance. The third installment will discuss LIBS analysis in some depth, exploring the various ways to go from a LIBS spectrum to a solution. Overall, this series is intended to provide an overview for those considering implementation of LIBS to solve a particular analytical problem, and an introduction for those interested in learning more about LIBS.

The growth of interest in laser-induced breakdown spectroscopy (LIBS) analysis of materials has been rapid during the past 10 years, similar to the rise of Raman spectroscopy or atomic emission spectroscopy in earlier decades. Part I of this series (1) documented the nearly exponential rise in LIBS papers from 1995 through 2010. We are now at the point at which many, if not most, people in the analytical community have heard of LIBS, and there are several companies offering commercial LIBS systems that are both laboratory- and industrially focused. The capabilities of the LIBS systems are defined first by their hardware, which defines the laser-sample interaction, the plasma, and the collected spectra, and second by the analytics that interpret the spectra. This installment of "Lasers and Optics Interface" focuses on the hardware portions of LIBS systems, particularly on the laser and spectrometer choices. 

Three basic parameters are at play in selecting and integrating a laser into a LIBS system: pulse duration, laser wavelength, and laser fluence on the sample. Secondarily, one may consider a double-pulse arrangement in which the ablation laser pulse and analytical pulse are separated by time on the order of microseconds. However, the double-pulse arrangement is typically deployed only in laboratory experiments or very specialized applications because of the added complexity and element- and matrix-specific enhancement observed in the LIBS signal. Because the double-pulse setup is not yet a feature in mainstream systems, we will leave out a detailed discussion of double-pulse interactions and refer readers to one of several excellent reviews available (2). 

Two basic choices are available for laser pulse duration, nanosecond-class lasers and femtosecond-class lasers. Characteristic lattice vibrations (phonon frequencies) are 10

  Hz, corresponding at the highest rates to wavelengths on the order of atomic spacing in solids. Laser pulses on the order of a nanosecond (10

  s) duration are thus very long with respect to characteristic lattice vibration periods in solids, so there is significant heat transfer in the solid during the laser pulse. Picosecond lasers, with pulses on the order of tens or hundreds of picoseconds, are typically in a similar mode of interaction to nanosecond lasers. It is only femtosecond lasers, with pulses typically on the order of 10

  s, that can interact with solid materials without significant heat transfer. 

The implications of the nanosecond versus femtosecond laser choice can be observed pictorially in Figure 1, which shows a laser ablation crater from each class of laser. The nanosecond laser crater (Figure 1a) looks like an impact crater on the moon, with a ridge around the laser impact point and debris scattered around the crater. The crater itself shows evidence of significant melting, and the debris are from direct physical removal and melting as well as nucleation and condensation of small particles from melted vapor in the laser plume. The femtosecond crater (Figure 1b), on the other hand, looks very smooth and appears to have no melting. Femtosecond ablation is much more of a mechanical process, with Coulomb explosion and direct photochemical bond breaking being a few of the major removal processes. Unlike nanosecond ablation spots, there is a minimal heat-affected zone because the process is not thermally driven. As a result, femtosecond ablation produces a more stoichiometric and representative mixture in the plume than does nanosecond laser ablation, because there are no thermal effects. 

 Figure 1: Craters obtained by (a) nanosecond and (b) femtosecond laser ablation. Adapted with permission from reference 3.			

For laser ablation inductively coupled plasma–mass spectrometry (ICP-MS) and laser ablation optical emission spectroscopy (OES), femtosecond lasers have been shown to provide more precision and accuracy than nanosecond lasers (4). This is thought to result largely because of the particle distribution from femtosecond ablation, which is markedly smaller (in average size) than the distribution from nanosecond ablation, and thus evaporates better in the plasma torch. However, for LIBS there are countervailing factors in favor of nanosecond lasers for LIBS. The plasma from a typical nanosecond laser is much more robust than a femtosecond pulse. As a result, the integrated emission may be greater, and studies have shown that early in the plasma electron densities and temperatures are higher (5). Overall, the detection limits are comparable for nanosecond and femtosecond LIBS, in most cases, and additional variables such as laser fluence and interactions with the particular solid material being analyzed play a role in the precision and accuracy of the measurement.  

Given the considerations in choosing between a nanosecond and a femtosecond laser, cost and reliability typically become the driving factors. Although great strides have been made in femtosecond laser technology in the past decade, femtosecond lasers are still roughly 8–10 times more expensive than comparable nanosecond-class lasers such as Nd:YAG. These lasers cannot be used outside of a laboratory because they are generally more fragile and subject to alignment and reliability issues than nanosecond lasers are. For these reasons, except for high-end laboratory systems that are primarily used in research and development, LIBS systems are typically outfitted with nanosecond-class lasers. 

Given a particular pulse duration for the LIBS system, the laser wavelength also has a large influence on the laser–material interaction that ensues. This influence is more pronounced for nanosecond lasers, because (as previously discussed) the pulse length is long enough to excite phonons in the material and interact with the plasma as the plasma forms. Hence, here I will focus on a few effects that are known to occur for nanosecond lasers. As we will see, the laser wavelength can influence both the ablation and plasma when using a nanosecond laser.  

A primary effect in the laser–material interaction is the laser absorption depth, which changes as a function of wavelength. As shown in Figure 2, for some materials such as silicon or germanium, the optical absorption depth at longer wavelengths is measured in tens or hundreds of micrometers or more. Such large characteristic absorption lengths mean that much of the laser pulse is absorbed in the depth of the material. In-depth absorption either causes large and unpredictable ablation, or the energy is deposited at depth and is lost, unable to participate in the ablation and plasma formation process. The rule of thumb that ensues from this consideration is that UV laser wavelengths — for example, 193-nm ArF excimer and 213- and 266-nm Nd:YAG — are required for the application of LIBS to materials that may be optically thin at longer wavelengths. Often, these are materials that appear transparent or translucent to the eye, but many polymers that appear opaque to the eye are still translucent at the fundamental 1064-nm wavelength of the Nd:YAG laser. 

Figure 2: Optical absorption depth as a function of wavelength. Data were obtained and adapted from reference 6.			

In general, shorter wavelengths are better for ablation using nanosecond lasers. Shorter wavelengths have smaller absorption depths in essentially all materials, which concentrates ablation energy and promotes efficient ablation. Also, the diffraction-limited spot size is proportional to the wavelength of the laser. As a result, short-wavelength lasers can be focused into smaller spot sizes, which is helpful for microanalysis.  

When it comes to plasma formation, however, there are some drawbacks to shorter wavelengths. As the electron density increases in the plasma, there is increased absorption in the plasma because of inverse Bremsstrahlung. The strength of the overall absorption can be described by the interaction of the incoming laser light and the rapid oscillations of electron density in the plasma known as "Langmuir waves." If the laser frequency is lower than the frequency of the Langmuir waves, known as the "plasma frequency," then there is a sharp increase in the absorption in the plasma. Fundamentally speaking, if the electrons are oscillating faster than the laser light, they can keep up with the electromagnetic wave and absorb it. If the laser frequency is higher, there is much lower absorption.  

Because the LIBS plasma is forming above a solid sample, the electron density increases from zero to a number on the order of 10

. Considering the evolution of the plasma during the 6–8 ns duration of a typical Nd:YAG laser pulse, the beginning portion of the pulse starts the ablation process, ejecting material into a plume and initiating plasma formation through other mechanisms such as multiphoton ionization. The latter portion of the pulse has the opportunity to interact with the expanding plume from the ablation. Depending on the wavelength of the laser, which for example could be 266 nm or 1064 nm for a Nd:YAG laser, the trailing portion of the pulse may be either more absorbed (in the case of a 1064-nm pulse) or less absorbed (in the case of a 266-nm pulse) in the plasma. A longer wavelength (lower frequency) laser will more efficiently couple in to the forming plasma and be absorbed by the plasma at an earlier stage in the plasma lifetime, corresponding with lower electron densities required for absorption. This improved coupling with longer-wavelength lasers results in a more robust plasma, with typically a larger volume and somewhat higher temperature and electron density. Hence, while the shorter-wavelength lasers have an advantage in ablation efficiency and spot size, the longer-wavelength lasers have an advantage in plasma formation. 

The total laser energy has an influence on both the material ablation and on the plasma. For a given sample, the amount of energy per unit area (laser fluence) is the relevant important parameter related to energy. It is the fluence that most directly influences the ablation volume. For some samples, increased fluence is too much of a good thing; for example, too much fluence in the analysis of gemstones can cause cracking or unsightly marks. Similarly, pressed pellets or tablets may crack or crumble if the fluence is too high. Therefore, it is not always important to have a high energy, and typically systems have adjustable energy and spot size so that the fluence can be tailored to the material being measured.  

Another interesting point is that at any given time the radius of the initial plasma is proportional to energy to the 1/5 power. In particular, 

 is time. The time, electron density, and temperature (

) progression can be shown to follow a roughly self-similar path (based on hydrodynamic theory) with 

				  (7). The upshot of this self-similarity in the decay time of the plasma is that the delay (after the laser) and the gate time (open shutter) of the LIBS detector can be adjusted in predictable ways when the energy is changed in a LIBS experiment. 

The detection systems used in LIBS can be parameterized by wavelength range, sensitivity, and speed and ability to gate the detector exposure. Overall, the number of choices and range of performance is continually expanding. Given the number of parameters, including cost, that are a consideration in a LIBS system, it's usually true that there is no one-size-fits-all magic bullet that is useful in every situation. Here, we'll cover the main points associated with each of the detector types to provide readers with an overall perspective to make an assessment.  

Charge-Coupled Device–Based Spectrometers

Used in many applications, charge-coupled device (CCD) spectrometers are inexpensive and widely available from many manufacturers. These spectrometers typically use a Czerny-Turner or crossed Czerny-Turner design. To be suitable for LIBS, they must be outfitted with a trigger so that the measurement can be initiated at a known time with no more than a 20–50 ns delay. Presently, none of the CCD systems on the market have a gating feature that allows the detection to be terminated at a known time; these systems are typically measuring for 1 ms or longer following the trigger. Detectors on the standard CCD systems (for example, Toshiba TCD-1304 or Sony ILX511B) are linear arrays and have a typical dynamic range of about 260–300. These chips or similar ones are used in all of the basic CCD systems, and thus all of them have a similar dynamic range. Different manufacturers use various analog-to-digital (A/D) converters with 12-, 14-, and 16-bit ranges, giving the appearance of additional sensitivity, but in reality the higher A/D ranges are simply window dressing and not performance-related. Spectrometers used in the UV region of the spectrum are typically enhanced with a lumogen coating. 

These spectrometers are arrangements of CCD spectrometers with a common triggering system. Accompanying software can be used to stitch together the output of these systems into a single broadband spectrum. At best, these systems generally provide 0.05 nm full width at half maximum (FWHM) resolution. 

Echelle spectrometers combine two dispersive elements, typically a prism and a grating, to disperse light in two dimensions on a square detector. This results in a two-dimensional spectral field with various orders in one dimension and wavelength in the other. Software is used to sort the various orders and construct a continuous spectrum, which can range from 200–900 nm or more. Light throughput in these spectrometers is not stellar, typically they have an effective F-number of 7 or 8. Depending on the wavelength of the emission line in question, echelle spectrometers coupled with intensified cameras (see below) can easily outperform CCD spectrometers by an order of magnitude or more. Resolution is typically 0.05 nm or better, but cost, particularly when considering the cost of the camera, can be substantial. 

These more traditional, large Czerny-Turner spectrometers may be 0.25 m in pathlength or more. Typically outfitted with a turret with several diffraction gratings, users can dial the center wavelength, wavelength range, and associated resolution to their liking. The light throughput on these systems is very high and light is generally dispersed using the (very strong) first order of the grating. On a typical intensified camera detector, the light is dispersed across the detector and users can choose to collect ("bin") the camera from one to the maximum number of rows on the camera. Because the gain on the camera can also be adjusted, this combination results in an instrument with incredible dynamic range (10

). The downside is that the limitations on the width of the detector result in the collection of only a narrow window of light (30–50 nm) with each shot, if 0.1 nm resolution or better in the spectrum is desired. Hence, associations between multiple elements may be difficult to obtain because the emission lines of the particular elements of interest may be in different parts of the spectrum. This limitation understood, the combination of components is one of the most sensitive LIBS detection schemes, surpassed only by the photomultiplier tube discussed below. 

Inexpensive CCD detectors were covered in some detail above. Beyond these very low-end detectors, there are numerous variations of CCD detectors that are more sensitive than the basic detector. Back-thinned detectors can have enhanced sensitivity, while high-quality CCDs that are cooled with multistage cooling can have signal-to-noise ratios that exceed those of an intensified CCD. The main drawback of these detectors is that the exposure cannot be gated (turned on and off) as can the other detectors. These are available in square (for example, 2048 × 2048) and "spectroscopy" (for example, 1340 × 100) formats.  

Intensified charge-coupled devices (ICCDs) have long been the mainstay for LIBS applications. As described above, ICCDs can be gated (some as fast as 2 ns) and have a dynamic range exceeding 10

. To accomplish this, the output of a photocathode is amplified (under vacuum) by a microchannel plate amplifier. The resulting electrons strike a phosphor, which is fiber-coupled to a CCD detector. The variable voltage that can be applied to the microchannel plate provides both the high gain potential and the gating capability of the device. Largely because of the uncertainties in the microchannel plate amplifier, ICCDs can be quite noisy at high gain settings. In addition, images are not as sharp as a typical high-quality CCD camera because there can be some "smearing" caused by crosstalk between the fibers coupling the light from the phosphor to the CCD. These cameras are quite expensive. 

Among the new kids on the block, electron multiplying charge-coupled device (EMCCD) cameras have generated some interest in the LIBS community because of their high sensitivity and much lower cost than ICCDs. In EMCCD cameras, the readout register has multiple preamplification stages before A/D conversion. This allows low-light signals to be enhanced before conversion, thus moving a low-light signal above the A/D noise floor. The gain possible in EMCCDs is about 10

. Precision and speed of gating, however, has been a development issue for LIBS. Because of the coupling of the preamplifier and the A/D conversion, the gating has an inherent minimum delay and the actual gate open time may vary across the chip because the chip is read out. Efforts are being made to address these issues to make EMCCDs more on par with ICCD gating performance. EMCCD cameras, as of this writing, are approximately 30–40% the cost of ICCDs, so they are worth watching as they mature. 

Photomultiplier tubes (PMTs) are detectors of the last century, correct? For the students reading this, these were the detectors that "old" guys like this author used with a scanning monochromator to obtain broadband emission spectra in 20 min flat. (Yes, one spectrum in 20 min. Today's students are spoiled!) Not so fast. With gains of up to 10

  and fast readout electronics available, in some situations they may be the most versatile detectors for LIBS. They can be used for either emission-line-selective filters or in a spectrometer configuration in conjunction with a Paschen-Runge spectrometer. Don't count these detectors out for the ultimate in speed (kilohertz acquisition possible) and sensitivity. 

Figure 3 summarizes the salient features of detection systems. Spectrometers or detectors are graded on a relative scale for broadband nature, sensitivity, triggering and gating, and cost. It is obvious, as we stated at the outset, that there are few clear winners. The one statement that can be made with confidence is that if one has a single, known emission line to measure, a PMT system is worthy of consideration. Presently, many OES systems still use PMT arrays for line detection for this reason. 

Figure 3: Relative performance of detection system components.			

This installment, part II in a three-part series, was intended to dissect the basic technology choices for components of LIBS hardware. There is not a perfect solution for all situations by any means, but there are excellent, and always improving, options for these components. We hope that this provides a basis to compare and contrast the laser, spectrometer, and detector choices such that when you are seeking a LIBS system solution from one of the manufacturers, you have a road map in mind. The last installment of this series will cover LIBS spectral analysis and the basics of proceeding from a spectrum to a solution. Stay tuned! 

(2) J. Scaffaldi, S.M. Angel, and D. Cremers, 

Columns: Lasers and Optics Interface

The choices for LIBS hardware are discussed in detail, particularly lasers and spectrometers, and the trade-offs between cost, size, and performance are illustrated.

LIBS Basics, Part II: Hardware

Issue PDF

Click the title above to open the Spectroscopy April 2014 regular issue, Vol 29 No 4, in an interactive PDF format.

Spectroscopy-04-01-2014

GLP and GMP Approaches to Method Validation — Going the Same Way?

Overview of High-Efficiency Transmission Gratings for Molecular Spectroscopy

Simultaneous Measurement of Chlorophyll and Water Content in Navel Orange Leaves Based on Hyperspectral Imaging

LIBS Basics, Part II: Hardware

Vol 29 No 4 Spectroscopy April 2014 Regular Issue PDF