Data Quality and Data Integrity Are the Same, Right? Wrong!

Do the terms data integrity and data quality mean the same? As “maybe” and “yes” are not the correct answers, we will explore the differences between the two terms, and how data integrity is a key part of data quality. 

	
	Data integrity has been the subject of many “Focus on Quality” columns over the past few years (1–5). However, we have never mentioned, let alone discussed, data quality. In many people’s minds, 

 are terms that have equivalent meanings, hence the first part of the title of this month’s column. However, in reading the MHRA (Medicines and Healthcare Products Regulatory Agency, the UK regulator) GXP data integrity guidance, there is an indication from a regulator that the two terms are different (6). Newton and White have discussed the differences between data quality and data integrity in an ISPE blog in 2015 (7), and, in this column, we expand and explore the similarities and differences between the two terms.	

		The MHRA’s 2018 guidance has in the introduction section the following statement in clause 2.7:

		“This guidance primarily addresses data integrity and not data quality since the controls required for integrity do not necessarily guarantee the quality of the data generated (6).”

		Clearly, in the mind of the regulator, there is a difference between data quality and data integrity. But what exactly is it?

		Our data quality versus data integrity story starts toward the end of the last century with ALCOA-not the Aluminium Company of America, but an acronym standing for 

attributable, legible, contemporaneous, original,

. These criteria have has been discussed in several data integrity guidances from the Food and Drug Administration (FDA), MHRA, the World Health Organization (WHO), and the Pharmaceutical Inspection Convention and Pharmaceutical Inspection Co-operation Scheme (PIC/S) (6,8–10). The WHO guidance contains the most detailed discussion of ALCOA criteria in an appendix with a definition of each term, presentation of the expectations for paper and electronic records, plus special risk management factors (9).

		A European Medicines Agency (EMA) publication on electronic source data extended ALCOA to include an additional four terms of 

 (11). The extension of these four additional terms is known as ALCOA plus, or ALCOA+. A recent article in 

 discusses all nine ALCOA+ criteria in more detail (12).

		Reading further through the MHRA guidance, there is clause 3.10, a part of which is presented below, discussing ALCOA and ALCOA+ criteria:

		“….. ALCOA was historically regarded as defining the attributes of data quality that are suitable for regulatory purposes…. [6]”

	This is interesting; ALCOA criteria were originally defined as the attributes of data quality? How did this come about? The ALCOA acronym was invented by an FDA good laboratory practice (GLP) inspector named Stan Wollen, who developed it as an aide memoire to remember the key requirements of data quality. This is outlined in a publication entitled “Data Quality and the Origin of ALCOA” (13), available for download on the web. There is a lot of background information together with Wollen’s interpretation of ALCOA under the US GLP regulations 21 

 58 (14). For our discussion, ALCOA was originally associated with data quality, as noted by the MHRA in its GXP guidance document (6), long before it was stolen by various guidance documents for use as a cornerstone of data integrity.

	Before we go into more detail about the two terms, we need to define them. The definitions for both 

 are taken from the 2018 MHRA GXP Guidance for Data Integrity (6), and are presented in Table I. We will consider and discuss the data integrity definition first, and then spend time looking at data quality. I am not going to discuss the last paragraph in the right-hand column of Table I about sound science and risk management, although those are key inputs to data integrity, but not part of the data integrity and data quality debate in this column.

	What is interesting about the 2018 definition of data integrity is how far it has changed since the first MHRA guidance published in 2015. The earlier definition was:

	“The extent to which all data are complete, consistent and accurate throughout the data lifecycle (15).”

	This definition is a modification of a 1995 FDA definition of data integrity (16). The 2015 definition can be compared with that published in 2018, where the additions are highlighted in bold text:

 to which data are complete, consistent, accurate, 

trustworthy, reliable and that these characteristics of the data are maintained 

	There is the addition of two criteria (trustworthy and reliable), as well as a requirement to maintain all five data integrity characteristics throughout the data life cycle.

	What is not stated in this definition, but in the explanation for raw data, is that dynamic data must be maintained in this format throughout the data life cycle. We shall return to this regulatory expectation, and the total inability of industry and suppliers to meet it, in a future “Focus on Quality” column.

As discussed above and in Table I, we have additional criteria listed in the MHRA definition: complete, consistent, trustworthy, and reliable, in addition to the five original ALCOA criteria (6). Two are ALCOA+ criteria that will be discussed below, plus two new requirements. ALCOA+ on steroids perhaps? Let us look at these four criteria, which to make matters worse are not defined in the guidance document:

 This is one of the ALCOA+ criteria, and is also an FDA 

 regulatory requirement in 21 CFR 211.194(a) where 

 are mentioned (17). Put simply, every item of data and metadata collected during an analysis from sampling to generation of the reportable result is covered under “complete,” including (but not limited to) any instrument log entries, documentation of mistakes, investigations, deviations, and instrument and calibration failures. Of course, you won’t be allowing any laboratory user deletion privileges, will you?

 Another of the ALCOA+ criteria that requires that the data be consistent with the processes being followed, and that the time and date stamps reflect the creation, modification, and audit trail entries are as expected. The process (manual or automatic, hybrid or electronic) should ensure that the acquisition, transformation, or calculation of data is transparent and traceable.

This is a new data integrity requirement that, together with “reliable,” harks back to the last century and the requirements outlined in 21 

 11 for electronic records and electronic signatures (18). In essence, can you trust the data or records? In some respects, it is the reverse of the ALCOA criteria. For example, if a record is attributed to an individual (ALCOA requirement), did that person actually do the work? Was the person in the laboratory at the time the work was performed, or did somebody act on their behalf?

 Another new data integrity requirement that means that all records and data are suitable and can be depended upon. Where a true copy has been made of an original record, is the verification process acceptable? The record set is not the 

th iteration of testing to pass and that work has actually been performed as required under EU GMP Chapter 1.8 and 1.9 for in process and quality control testing, respectively (19)?

	The addition of trustworthy and reliable to the ALCOA+ criteria are good. So, what does this mean for data integrity as a whole?

	Data integrity provides assurance that the analytical work in the laboratory from the sampling to the calculation of the reportable result has been carried out correctly. In short, can you trust and rely on the analysis and the reportable results? If you cannot trust the analysis and the data, then you cannot make a quality decision based on poor, incomplete, or falsified data.

	It is the role of the analytical scientist and the second-person reviewer to ensure data integrity, but that also requires that the instruments, computer systems, analytical procedures, and laboratory environment are set up correctly, and that the organization within which they work ensures that they are not pressured to cut corners or falsify data.

	Therefore, ensuring the integrity of data is a major contributor to data quality. But, in focusing on data integrity, we must not lose sight of data quality.

in Table I that there are three main elements: Data produced is exactly what was intended to be produced       It is fit for its intended purpose

	Hmmmm, what do these mean in practice? Let us look at item 1, that data produced is exactly what was intended to be produced. The problem with this is the wording of the definition. Although a laboratory produces mountains of data, we abstract and use the information within them to make decisions. For a detailed discussion of this process together with the controlled and uncontrolled parameters, see the article by McDowall and Burgess (20). Therefore, the purpose of a laboratory is to generate information from the analysis of samples in order to make decisions. These decisions could be to aid product development or release a batch of product.

	To do this requires trained staff, qualified analytical instrumentation, validated software, and analytical procedures verified or validated under actual conditions of use, as required by 21 

 211.194(a)(2) (17). This is the same as the first three levels of the data integrity model discussed in earlier articles (21–23). Where the second item comes in is that the output of the analytical procedure must be fit for use. As a simple example, there is no point calculating results of an analysis to three significant figures if all that was required was the presence or absence of the analyte.

	Once you have ensured the integrity of data, we must consider data quality. In part, this focuses on the laboratory’s ability to deliver the results to the sample submitter and meet their requirements for the analysis such as:

		Precision and accuracy or measurement uncertainty of the result

		Supporting information for the analysis

		Speed of analysis (for example, for releasing a production batch or turnaround of analysis from a clinical trial)

	However, there is a compromise with data quality that is best illustrated by the data quality triangle.

There is a problem with the definitions of 

 that is not mentioned by the MHRA and the American Health Information Management Association (AHIMA); this is the data quality triangle, as shown in Figure 1. At each end of the triangle are the following attributes:

		Cost: cost in instrumental and human resources

		Quality: delivering data of acceptable quality to the end user of the information to make a decision.

	The problem is that you can only select two of the three at any one time. For example, if you want a fixed time and high quality, how much are you willing to pay or how much resource will you commit? Equally, if you want analysis performed quickly and at low cost, quality suffers. The quality triangle is applicable to any laboratory in any industry, but in a regulated laboratory you still must ensure data integrity as well. The influence of senior and laboratory management on managing the three criteria of the data quality triangle will directly impact both data quality and data integrity.

Integrity and Quality Use the Same Data Set

	It is imperative to realize, as mentioned earlier, that the data set for ensuring data quality and data integrity is the same, and not two different ones. Furthermore, data integrity is not ensured first and followed by a second pass to ensure data quality. To ensure data quality and data integrity, both must occur in parallel, simultaneously and on the same dataset.

	Because the two terms are different, as Newton and White (7) point out, you can have data integrity without data quality, and vice versa. A laboratory can have excellent data integrity, but if the analytical information to make a decision is delivered late, or not at all, then the data quality is useless. Alternatively, if each laboratory is permitted to create its local practices for data management, you can have integrity without quality (or, at least, very low levels of quality). All the data are there, but consolidating them requires a standardization project: good integrity, lousy quality. That is the one place where quality and integrity do differ, and where the regulations are weak (Newton, personal communication).

	Newton and White (7) present a case where an organization outsources analytical work to contract laboratories, but there is little oversight of the work, resulting in data integrity lapses. When the results are received by the organization, entry to the LIMS controlled and delivery of the results to the decision makers is rapid, and with good data quality.

	The similarities and differences between data quality and data integrity are shown in Figure 2. The similarities are:

		Application of ALCOA and ALCOA+ criteria to the record set

		The differences between the two terms are trusting the data versus making a decision using the results.

	The problem and potential conflict comes with the third element of the MHRA definition for data quality that states that data quality includes ALCOA. This is consistent with Wollen’s original view when devising the acronym ALCOA (13). However, when we see the MHRA definition of 

 in Table I, each ALCOA criterion is spelled out in the definition. Puzzled?

	Is ALCOA an integral part of data integrity, which in turn is a component of data quality, or is ALCOA applicable to both terms, as noted in the MHRA definition?

	Look at it from the perspective of the data and records being generated. Can you separate the data for data quality from data integrity? The answer is no, because the record set for both data integrity and data quality will be the same. Therefore, ALCOA criteria apply equally to data quality and data integrity. Think it through; otherwise you could have some data where all actions were attributable for data integrity, but not for data quality.

	Figure 2 shows the relationship diagrammatically; in the middle is the analysis workflow from sampling to a decision based on the reportable result from the analysis. Throughout this process, data and records are generated that comprise the complete data from the analysis. This record set is impacted by the ALCOA criteria (as well as the four additional requirements) for data integrity mentioned in the MHRA definition in Table I and the ALCOA criteria for data quality.

Articles

Focus on Quality

Data integrity has been the subject of many “Focus on Quality” columns over the past few years (1–5). However, we have never mentioned, let alone discussed, data quality. 

 is a collection of column articles that the authors published in 

 over a period spanning more than two decades. Each article is generally arranged as a chapter in the book, and chapters dealing with the same or similar topics are arranged closely as a section block rather than following the original sequence in the magazine. Although each article or series of articles only discusses one specific topic, collectively, the articles form a comprehensive reference that is a valuable source for readers wanting to learn chemometrics, especially with its applications in spectroscopy. 

	The book is divided into 128 chapters, split over 24 sections, and a vast 1040 pages. Some might ask whether it is really necessary to wade through such a weighty tome to learn chemometrics. The answer could be “no,” with the argument that for the four multivariate methods delineated in the book, multiple linear regression (MLR), principal component regression (PCR), partial least squares (PLS) regression, and classical least squares (CLS), each could be described in no more than perhaps two pages in the matrix form. Furthermore, user-friendly software packages are readily available and generally easy to use, and in many cases it might only take a few mouse clicks to build a calibration model. At the end, if the predictions from the calibration model show marked differences from the actual data, one can apply the famous “garbage in, garbage out” principle to declare the data garbage, which is not uncommon to see nowadays in real practice.

	However, if one believes that good data should not be discarded as garbage without good reason, especially when months, or even years, of effort may have been invested to collect them, it is then worthwhile to read this book, particularly the sections about CLS. Readers will find that even very experienced chemometricians could not develop the right model the first time, simply because the wrong unit of measure was chosen for concentration. CLS is the simplest method among all the methods presented in the book, and could easily be described by two or three equations. However, the authors spend 14 chapters discussing the method and its applications in considerable length. The authors could have simply told the readers that volume percent or a comparable unit should be used as the concentration unit rather than weight percent. Instead, the authors spend many pages describing their troubleshooting process along with the identification and verification of the root cause. Readers will perhaps benefit more by reading the lengthy description of the problem-solving process, rather than learning the simple reason in one sentence.

	One cannot avoid mathematics when learning chemometrics. Mathematics is needed to understand the equations in the multivariate methods and also many statistical methods for evaluating raw data and model predictions. A beginner does not need much mathematics background to learn chemometrics using this book, however. Matrix algebra is introduced in Section 1, and then further expanded in Section 2, together with the description of MLR. Analytical geometry, which is needed for the understanding of regressions, is introduced in Section 4. The rather detailed description of the mathematical methods using simple language may seem somewhat tedious to an experienced reader, but is certainly beneficial to someone who is less comfortable with equations. PCR and PLS are introduced similarly by giving the detailed calculations using the simplest possible data set. The concept of principal components is perhaps the most difficult to understand for people who are new to multivariate analysis, or do not have a strong background in matrix algebra. To help readers understand principal components, the authors present an approach that uses only elementary algebra to derive the algorithm. This approach is long, complicated, and rather tedious, but could enable readers who are new to this topic to understand every single step intuitively.

	Statistics is a very big area for chemometrics, with many methods and books available. The authors have very strong backgrounds in statistics, and have certainly included necessary and adequate statistical concepts or methods for chemometricians. Design of experiments (DoE) or experimental design, including analysis of variance (ANOVA) is introduced in Section 3 using easy-to-understand language. DoE is a very relevant topic, and the introduction is rather high level for such a significant area. Section 7, titled “Collaborative Laboratory Studies,” contains a detailed description of statistical methods for comparing different analytical results and methods, although these methods can also be found in regular analytical chemistry books. Section 10, “Goodness of Fit Statistics” describes statistical tools needed for analyzing linear regression. Section 20, “Statistics,” contains articles introducing the three foundations of the subject, which are useful for people without much statistical background. Useful discussions of statistics can also be found in Section 12 (“Connecting Chemometrics to Statistics”) and Section 15 (“Clinical Data Reporting”).

	Multivariate methods such as PLS can be, and have been, used on data from various scientific disciplines when the data can be arranged in formats that are suitable as inputs and outputs of the methods. The multivariate methods perform the same calculations regardless of the source of the data, whether the data are from chemistry or economics. However, the methods for preprocessing the raw data to render them suitable for the multivariate methods to yield the best prediction model can be very different, depending on the data source. The data and examples used in this book are mainly related to near-infrared (NIR) or mid-infrared (mid-IR) spectra. There are no discussions of infrared spectrometers or spectroscopic measurements in the book, perhaps because the authors assumed that the readers of 

 already had the necessary knowledge or experience. Readers who are unfamiliar with IR spectroscopy would need to obtain some basic knowledge through other sources to get the most benefit from this book. 

	The mathematics described in this book are very basic, such as Section 1, “Elementary Matrix Algebra.” Conversely, the chemometrics-related discussions on spectroscopic data are rather advanced, a reflection of the knowledge and experience that the authors possess through many years of practicing and writing in this field. The book contains deep discussions of derivatives, linearity, noise, and outliers, all of which are critical for building good chemometric models.

	One very important area, perhaps as important as building the original calibration model, is calibration transfer. IR spectroscopic methods should be able to replace some traditional wet chemistry methods, such as chromatographic methods that consume solvents and generate significant volumes of waste, something that is truly needed in the green chemistry era. Many times it was the maintenance of a calibration model, rather than the initial cost of building the model, that prevented the acceptance of the spectroscopic methods within commercial manufacturing environments. Transferring calibration models is not impossible from a technical standpoint but the required knowledge of chemometrics is generally not sufficiently widespread. Calibration transfer is well covered in three sections or ten chapters in the book. All aspects of calibration transfer, such as reference standards, instrument performance, and modeling algorithms are described in the book. A review of published methods on calibration transfer is also included. 

	Since the book is written in column format from the authors’ own columns in 

, it contains lots of authors’ comments. These comments were not always agreed upon by other experts in this field. As an example, the initial article on linearity in Section 6 included discussions with several other experts, when a simple noise-free synthetic nonlinear set of spectra was used to compare MLR, PCR, and PLS. The discussions reveal that, even with such a very simple set of spectra of a single component, expert chemometricians could use different approaches and interpret the results differently. Building a multivariate calibration model comprises a fixed sequence of steps, but there are various approaches that can be taken at each step. The combinations of these approaches across all steps can be numerous, resulting in many different models. Reading these discussions may inspire more efforts on building better calibration models rather than concluding that the raw data are garbage!

	Some improvements can perhaps still be made in future versions or in the book’s companion website, a publisher-hosted resource where the authors can add additional material in electronic format. Color figures are already available on the companion website if they were printed as black-and-white in the hard copy. Since the book was not written as a textbook by first intent, it does not provide example problems for the readers to practice with. The book does provide data or scripts that are in MATLAB or Mathcad format, however. Maybe in future editions or on the companion website, the authors can add Python or R versions of the data and scripts to benefit readers who do not have access to the named software packages.

	For such a large book that encompasses so many diverse topics, it would help readers navigate the information presented if the titles of some chapters, although not many, could be updated to more closely reflect their contents and made available on the book’s companion website. For example, the book contains 15 chapters in Section 8, “Analysis of Noise.” Many types of noise and their effects on spectra are discussed, together with sophisticated equation derivations. However, the chapters themselves are simply named Part 1, Part 2, and so on, up to Part 15. It would be very helpful to readers if the titles indicated which type of noise is discussed in each chapter.

	Some minor topics or discussions are likely to be of particular interest to specific groups of readers, but, unfortunately, they cannot be found either from the table of contents or from the index. Aside from Section 15, “Clinical Data Reporting,” there are three additional chapters that are particularly useful to users in the pharmaceutical industry. However, readers would not know of their existence unless they happen upon them. Some readers might be comfortable to read the book sequentially using the current arrangement of the chapters, but other readers will need to read the chapters in a different order, or will only need to read some selected chapters. It might be useful if the authors can provide some study guides, tailored to different reader groups, and add them to the book’s companion website.

	Neural network (NN) is one area that is mentioned, but not discussed, in the book. NN is a useful method when dealing with nonlinear data. Similarly, multivariate curve resolution (MCR) has found increased and successful use in recent years, but is not included in the book. MCR has the potential to remove the reliance on a primary analytical method for certain types of quantitative analysis. Both techniques are worth adding to future editions of the book. Classification of spectra may also need to be expanded in future editions.

	This book certainly contains sufficient material for beginners to learn chemometrics and for advanced users to strengthen their skills for dealing with more challenging calibration problems or even to create innovative applications in vibrational spectroscopy. Written originally in journal column format, the book contains lots of author’s comments, and very detailed explanations (in the author’s words, “to treat any topic at whatever length is necessary”). Whether readers like the book or not may depend on whether the reader enjoys the writing style of the column format, and if the reader can quickly locate topics of interest. Readers might want to obtain some sample chapters, which should be possible through the publisher or past issues of 

, to find out if they are comfortable with the writing style. An enhanced table of contents, or some type of study guides, would help readers find their desired contents faster.

	Husheng Yang is a senior scientific investigator and chemometrician at GlaxoSmithKline, in  Collegeville, Pennsylvania. The reviewer is not acting as a representative or agent of GSK. Direct correspondence to: husheng.x.yang@gsk.com

Book Review

Chemometrics in Spectroscopy is a collection of column articles that the authors published in Spectroscopy over a period spanning more than two decades. Each article is generally arranged as a chapter in the book, and chapters dealing with the same or similar topics are arranged closely as a section block rather than following the original sequence in the magazine. Although each article or series of articles only discusses one specific topic, collectively, the articles form a comprehensive reference that is a valuable source for readers wanting to learn chemometrics, especially with its applications in spectroscopy. 


Book Review: Chemometrics in Spectroscopy, 2nd Edition, by Howard Mark and Jerry Workman, Jr.

Spectroscopy, November 2019

Click the title above to open the Spectroscopy November regular issue, Volume 34, Issue 11, in an interactive PDF format.

Vol 34 No 11 Spectroscopy November 2019 Regular Issue PDF

Raman spectra of five reference samples of ethylene vinyl acetate (EVA) with vinyl acetate (VA) copolymer composition between 14 and 40% were measured. After normalizing the spectra to the >CH

, the spectra can be compared and band assignments made, enabling a description of morphological changes resulting from the introduction of the vinyl acetate group. The spectra were then analyzed by two-dimensional correlation spectroscopy (2D-COS). The 2D-COS enabled an easy way to use the spectral data to derive information on conformational changes that bypasses band-fitting, which can be non-unique.   

	
	Polymers are experiencing ever-expanding uses in many applications. The successful use of polymers depends on their chemical and physical properties which, in turn, depend on the molecular weight, chemical composition, and morphology. When a polymer is used to dissolve a small molecule (for instance, for drug delivery or for pigmentation), it needs to be noncrystalline. When it is used for a structural application, it needs to be tough, to be capable of resisting fracture when stressed. For those who are new to the characterization of polymers, it is useful to explain that 

 refers to the special arrangement of amorphous versus crystalline composition, orientation of the molecular chains, and the state of amorphous chains in direct vicinity of the crystalline phase, all of which affect the physical properties.

	In my journey into acquiring expertise in two-dimensional correlation spectroscopy (2D-COS), I have been looking for a system where I can connect what I know about the spectra with the changes that 2D-COS is showing. Some of the work that I have published in this column on the spectra of polyethylene terephthalate was designed with this in mind. However, this system is quite complicated, so I have been looking for a  system where we can more easily use the combination of Raman spectroscopy and 2D-COS to reveal useful information about chemical or conformational changes. After I collected spectra of five reference samples of poly(ethylene vinyl acetate) (EVA), I dropped them into 2D-COS and asked my mentor, Isao Noda, if this would be a convenient approach to try. His enthusiastic response was my go-ahead signal. For me, this is exciting, because I am able to rigorously follow the physical and chemical changes in a material. The focus is entirely different from my usual focus on microanalysis. Raman spectroscopy has not been used as much for chemical characterization as infrared (IR) and nuclear magnetic resonance (NMR) spectroscopy, for a variety of reasons. Its application requires detailed knowledge of spectral interpretation, which is not being taught as it was 50 years ago. For me, this has provided an incentive to learn more about spectral interpretation, especially of organic compounds.

	Ethylene vinyl acetate is a copolymer of ethylene and vinyl acetate with the latter usually present at concentrations less than 50%. Whereas polyethylene (PE) can be quite well-ordered (chains parallel with long-range crystalline packing order), the introduction of vinyl acetate interrupts that order. The spectrum of polyethylene has been extensively interpreted (1), so the changes introduced by the presence of vinyl acetate are amenable to interpretation. The first figure shows the Raman spectrum of poly(vinyl acetate) (PVA, top), EVA, 40% vinyl acetate (middle), and PE (bottom). Before examining the spectra, one might expect that the spectrum of EVA will be a superposition of PVA and polyethylene, but, in fact, that is an extreme oversimplification that becomes clearer by examining the spectra of solid and molten high density polyethylene (HDPE) in Figure 2. The top of the figure shows the spectrum of a pellet of polyethylene (2) at room temperature where it is solid versus the bottom spectrum of the same material, recorded at 200 â°C where it has melted. In the case of EVA, as the concentration of the VA is increased, the crystalline order of the polyethylene content is interrupted. Disorder in polyethylene has well-known effects on the spectrum, which can be seen in Figure 2. In molten polyethylene, there are three broad bands in the fingerprint region-the C-C single bond at about 1070 cm

. If polyethylene has a limited number of side branches, when solidified, the chains will stretch out with translational alignment; there are two methylene units in the repeat segment of the chain, and two chains in the unit cell. Using this information, it becomes possible to assign molecular motion to all the peaks in the spectrum of the solid (1,3). With this information in mind, when we examine the spectra of the various compositions of EVA, we note that bands of well-ordered, crystalline polyethylene are decreasing as the VA composition is increasing, and VA bands are increasing with VA concentrations. What the 2D-COS will tell us is what is happening simultaneously (in the synchronous plots), and what is not happening simultaneously (in the asynchronous plots). What is even more important is 2D-COS will tell us the order in which the different bands are changing. Before thinking this through thoroughly, it did not occur to me that the conformational changes would not be simultaneous. This has important implications for the study of temperature dependence and extrusion in polymers.

	In this section, I enumerate the bands that I am using to follow  the influence of VA on the polyethylene order. Figure 2 shows the Raman spectrum of a pellet measured just above room temperature (30 â°C), and then at 200 â°C, which is well above the range of melting temperatures (120–180 â°C). In the liquid phase, there are three broad and prominent bands in the fingerprint region-the C-C stretch near 1070 cm

. Polyethylene is an unusual polymer in that it can exist in a highly crystalline state. In that form, the C-C stretch splits into two sharp bands at 1060 and 1125 cm

, the backbone stretch is sharp and appears near 1295 cm

 deformation region exhibits sharp bands near 1420, 1440, and 1460 cm

. In the CH stretching region, crystalline polyethylene is dominated by a strong sharp band near 2880 cm

, with a second, somewhat broader band at about 2840 cm

, with about 65% of the peak intensity of the first band. As disorder increases, the band at 2880 cm

 band, and higher frequency shoulders grow in intensity and sharpen. These are the features that we will be following in the 2D-COS behavior.

	The spectra of the EVA samples in the fingerprint region are overlaid in Figure 3a, after scaling all the spectra to the intensity at the deformation near 1440 cm

. Note that in this region there will be contributions from the methylene groups in the polyethylene segments, as well as the methyl group in vinyl acetate, and this may introduce some non-linearity to the spectra, but we will use the spectra scaled in this manner for this work, because we are looking more for effects of loss of crystallinity rather than small chemical changes. To make the examination of the spectra easier, I have color-coded the samples from violet to blue to green to red to brown as the concentration of VA changes from 14% to 18% to 25% to 33% and then 40%. The peak labels are brown (when the VA is at high concentration) or violet (when the VA is at low concentration). Thus, the violet labeled peaks correspond to peaks from the crystalline polyethylene units, and the brown labeled peaks correspond to peaks from the VA, and this can be confirmed by comparing these spectra from the various figures to those in Figure 1. Figure 3b shows the same spectra but in the CH stretching region. (The sharp band near 3060 cm

 is an artifact from the room lights and should be ignored.) The purple and brown color-coding is identical to that of Figure 2a, but the apparent peak at 2903 cm

 is highest in the 30% sample, not 14 or 40%, so it is labeled in the color of that sample. What will become apparent is that this intensity at 2903 cm

 is not due to a real peak, but the overlap of bands in the two phases that happen to maximize in this sample.

2D-COS-Following Conformational and Morphological Changes

	My goal is to understand how the conformation of the crystalline phase of polyethylene is changing as the VA concentration increases, and, in particular, what is the order of the changes. Before learning   about 2D-COS, my assumption was that as a polymer changed, all parts of the material changed together. If this were the case, the asynchronous plots would be blank. Now we are able to determine the order of changes in the polymer as reflected by the order of changes in the spectral bands. In particular, we will follow bands of the crystalline versus amorphous forms of polyethylene, keeping in mind that the intensity of the marker bands for the crystalline phase go down, and the amorphous bands go up, as the VA is added. In addition, it is important to keep in mind that the intensity of the spots in the 2D-COS plots indicate the magnitude of the changes, not of the peaks themselves. In fact, in the diagonal points of the synchronous plots, all peaks have the same sign. That means that while the peaks indicating a disordered polyethylene lattice are increasing, those of the crystalline lattice are decreasing, but they are changing together. It is only by examining the off-diagonal points that we can determine which peaks are increasing, and which are decreasing. But to determine the sequence of the changes, we have to examine the asynchronous plots.

	If I plot the entire 2D-COS spectra, it will not be possible to see the details of interest. I will be plotting first blocks along the diagonal, then blocks off the diagonal, to examine the behavior of cross-peaks. For example, I know that the sharp CH peaks near 2840 and 2880 cm

 are attributable to the crystalline phase. If I want to understand how the changes in the CH region line up with changes in the CH

 deformation or the backbone stretch or the C-C stretches, then I will produce the 2D-COS plots with the CH region on one axis and a fingerprint region on the other axis.

	Figures 4 through 9 show the 2D-COS results in the following regions:

	The rules for determining whether the change at ν

, or visa versa (4) are shown in Table I. Table II shows the sequence in which the bands are changing that were derived from the 2D-COS plots. In order to produce this table, one has to inspect each of the partial 2D-COS plots and determine the sequence of the changes for the bands of interest in that plot. Then, when all the plots have been evaluated, it is possible to consolidate the results and produce the list shown in Table II. However, it is a bit tedious to get everything in a self-consistent order; it is more or less like solving a puzzle.

	So what does this list mean? The first bands listed are the two VA bands near 1350 and 1370 cm

. Although polyethylene does have a band in this region (the so-called 

 groups), most of the change here is from the VA. In fact, careful examination of the overlaid spectra in this region (Figure 3a) does indicate that at low VA concentration, the band at 1368 cm

 is sharper, and more intense, than the band at 1346 cm

, but at high VA concentration the two have similar widths and intensities exactly as in the VA and 44% EVA as shown in Figure 1. The next band to exhibit changes is the backbone twist in amorphous polyethylene, which, of course, grows in intensity with VA; note that this band is totally absent in PVA, so its presence indicates only the amorphous content of polyethylene, and no contribution from VA.

	Following this are changes around the >CH

 deformation that was used for normalization at 1437 cm

. What is curious is that the two small spots that seem to indicate symmetric increases in intensity at 1427 and 1440 cm

 in the synchronous plot do not behave in parallel. The 1440 cm

 band of crystalline polyethylene decreases next, and then the 1425 cm

 spot increases. To convince myself that this conclusion really reflects the data, I have plotted this region of the spectra in Figure 10. The figure shows clearly the decrease in the crystal bands at 1415 and 1460 cm

, but it also shows broadening around the centroid at 1437 cm

. The spots appear in the 2D-COS plots because of the -CH

 group on the acetyl that replaces one of the protons on the chain. In fact, the normalization to the centroid of the >CH

 probably enhances the ability to see the increase in width. The spectral plot does not indicate directly the order of the changes, but at least it does support the conclusions from the 2D-COS.

	Following these changes, there is the growth of the amorphous contributions in the CH region at 2932 cm

. Then there is the loss in intensity of the crystalline CH bands near 2880 and 2840 cm

. After that there is growth in the amorphous C-C stretch at 1080 cm

, followed by decrease in the crystalline C-C stretches at 1060 and 1125 cm

. Finally, there is loss in intensity of the backbone twist of the crystalline polyethylene, and then the loss in intensity of the 1460 cm

	When I examine these changes, the trend that I see is that first we observe growth in the VA bands that have no overlap with polyethylene (1346 and 1368 cm

), and then the changes seem to happen with the growth of the amorphous phase before the loss of crystallinity. In most cases, changes of analogous bands (amorphous and crystalline) occur sequentially; see the bands in the CH region and then the C-C region. The last bands to change are the backbone twist of the crystalline phase, and then the last component of the >CH

 deformation region puzzling. While both the 1415 and 1460 cm

 are clearly associated with crystallinity, they did not behave the same. But the following will explain what is going on. In the amorphous phase there is a single chain δ(>CH

symmetry. In the crystal there are two chains per unit cell, and polarization studies have shown that this mode splits into bands at 1415 and 1440 cm

band in the crystalline phase, the overtone of the IR active band is invoked. The single chain >CH

in the amorphous phase, but splits into a doublet at 720 and 730 cm

 in the crystalline phase. Thus the overtone of the 720 cm

 component of the split-off Raman band, and the overtone of the second at 730 cm

 will be the Raman band observed at 1460 cm

. These assignments have been summarized in (1).

	It is tempting to try to rationalize the order in which the band changes occur. That the first change was the introduction of VA bands is easy to understand. But why would the appearance of the backbone twist of the amorphous phase appear so much earlier than the disappearance of the twist in the crystalline phase? Why does the disappearance of the 1415 cm

 crystalline band occur between the two spots that appear in the 2D-COS at 1440 and 1425 cm

; perhaps because the magnitude of the changes on the two sides of the 1437 cm

 band are not equal? One can argue that the changes in the CH stretches precede those of the CC stretches and crystal backbone because it is easier for the protons to adjust to changes than the CC units, which are constrained and coupled together. Likewise, it can be argued that it is easier for there to be adjustments in the single CC bond stretches than the backbone twist, which probably requires some minimum coherent sequence length along the chain.

	After this work was completed experimentally and I was preparing to write, I picked up the text by Noda and Ozaki and it fell open to pages in which analysis of this same chemical system was described (5)! The work cited in this section of the text goes back to 1999, in which Fourier transfer (FT) Raman spectra were analyzed only in the fingerprint region. The results were discussed in terms of the three-phase model (the orthorhombic crystal phase, a melt-like amorphous phase, and an isotropic disordered phase). Some of their conclusions are different than ours, but they did not describe how the spectra were normalized for comparison. I believe that the analysis and conclusions can be different, because I have normalized spectra at 1437 cm

	Of what use is all this? The question is important, because in deriving Table II, much time was involved to check and recheck the sequences and I have to ask myself if it was worth doing. Because polymers are used for many applications with quite differing physical and chemical requirements, and the physical and chemical properties depend not only on their chemical composition, but on their history, this type of analysis can provide information that is difficult to acquire by other means. In particular, polymers are often extruded from the melt and the speed with which they are extruded as well as the temperature and other factors will determine the physical properties, including (but not limited to) orientation, crystallinity, and ductility. Because the 2D-COS results provide information on the atomic/molecular scale, the polymer chemist can access this information to engineer their materials. For instance, the ductility of a polymer depends (inversely) on the crystallinity. If the EVA is being made for an adhesive application, it will need to be ductile, but not so much that it will flow. Using the information provided in Table II, along with MVA (multivariate analysis predictions), one should be able to determine the morphological characteristics effecting the ductility, and then adjust the manufacturing conditions for optimization. For example, tuning the extrusion temperature, or  amount of shear, could potentially change the order in which these changes occur, which could affect the bulk physical properties.

The Vibrational Spectroscopy of Polymers (

Cambridge University Press, Cambridge, United Kingdom, 1989).

		Scientific Polymer Products, Polyethylene High Density Pellets CAS#25213-0209 Cat#041, Ontario, NY.

Two-Dimensional Correlation Spectroscopy 

(John Wiley and Sons, Ltd. Chichester, United Kingdom, 2004).

		Section 10.4 Composition-based 2D Raman Study of EVA Compounds in Reference 4.

is the Principal Raman Applications Scientist for Horiba Scientific in Edison, New Jersey. Direct correspondence to: 

Raman 2D-COS spectral data provide information on conformational changes of polymers. Here, Raman spectra of ethylene vinyl acetate and vinyl acetate copolymer are measured and interpreted, enabling a description of morphological changes related to the vinyl acetate group.

Raman Analysis of Ethylene Vinyl Acetate Copolymers–Using 2D-COS for Identifying Structural Changes

Amides are an important functional group, because they are found extensively in polymers and proteins. They are unusual in that they contain nitrogen and a carbonyl group, giving a number of useful group wavenumbers. Also, hydrogen bonding affects their spectra.  

	
	The amide functional group is important in biology and industry; every protein molecule on the planet contains amide linkages, and the backbones of the nylon family of polymers contain amide bonds. Amides are consistent with the theme of the last several columns in that they contain a nitrogen atom, but they also contain a carbonyl or C=O bond, which we have studied extensively in previous columns (1). The generic structural framework of amides is seen in Figure 1. 	 	

	The first thing we have to decide is how to pronounce the word “amide.” Depending upon who I am talking to and where they are from, I have heard about nine different pronunciations for this word that, when spelled out phonetically, are: Ay-mids, Ay-mides, Ay-muds, Uh-mids, Uh-mides, Uh-muds, Am-mids, Am-mides, and Am-muds. My preferred pronunciation is “Ay-mides”, which I will use throughout the rest of this column.

	The amide functional group consists of a central carbonyl group with a nitrogen atom single bonded to the carbonyl carbon. This nitrogen is called the “amide nitrogen”, and can have carbons or nitrogens attached to it. When we first started studying organic nitrogen compounds and were introduced to amines, we found that they came in three varieties, called primary, secondary, and tertiary, depending upon the number of carbons attached to the nitrogen atom (2). Similarly we speak of primary, secondary, and tertiary amides, and again the difference between them is the number of carbons attached to the nitrogen. The skeletal frameworks of primary, secondary, and tertiary amides are seen in Figure 2.

	Note that, as in amines, we are counting the number of C-N bonds. Thus, a primary amide has one C-N bond and two N-H’s, a secondary amide has two C-N bonds and one N-H, and a tertiary amide has three C-N bonds and no N-H bonds. Given that this primary/secondary/tertiary terminology applies to both amines and amides, we can speak more generally of primary, secondary, and tertiary nitrogens. Therefore, when looking at the structures of amides, amines, and other nitrogen containing functional groups, a primary nitrogen will have one C-N and two N-Hs, a secondary nitrogen will have two C-N bonds and one N-H, and a tertiary nitrogen will have three C-Ns and no N-H bonds.

	In previous columns when we studied carbonyl groups attached to benzene rings, we encountered the phenomenon of conjugation (3). Briefly, the pi orbital on the carbonyl group overlaps slightly with the pi electron cloud orbital on the aromatic ring, causing a small amount of electron density to be withdrawn from the carbonyl bond lowering its force constant. Spectroscopically, this is expressed as a ~30 cm

 lowering of the position of the C=O stretch compared to a non-conjugated carbonyl (3), and we spoke of 

 versions of carbonyl containing functional groups (3).

	Amides also engage in conjugation as seen in Figure 3.

	The nitrogen atom in amides contains a p-orbital with a lone pair of electrons in it. This orbital happens to point in space towards the pi-electron cloud of the carbonyl group. There is some orbital overlap here, as illustrated in Figure 3, which leads to conjugation. Like with aromatic carbonyl groups, as a result of conjugation some of the electron density is withdrawn from the carbonyl bond, weakening it and lowering its force constant. This causes the amide C=O stretch to be on the low side compared to other carbonyl containing functional groups, typically from 1680-1630 cm

. Note, however, that we do not speak of 

 amides as conjugation takes place in all amides so they all have the same carbonyl stretching peak range.

	When we first started studying the infrared spectra of organic nitrogen compounds, we saw how hydrogen bonded to each other (4). We also discussed at length how N-H stretching peaks appear in the same wavenumber region as O-H stretching peaks, but that N-H stretching peaks are weaker and narrower than O-H stretching peaks, because the hydrogen bond strength for N-H is less than that for O-H (4). We then saw how amines hydrogen bond to each other, and what N-H stretching peaks look like (2,5). Like amines, amides also engage in hydrogen bonding, but it is structurally different than in amines, as seen in Figure 4.

	Because of the electronegativity difference between carbon and oxygen, the oxygen atom in a C=O bond contains a partial negative charge, as denoted by the Ð±- in Figure 4. Because of the electronegativity difference between nitrogen and hydrogen, the hydrogen in an N-H bond has a partial positive charge, as denoted by Ð±+ in Figure 4. The result is that the carbonyl on one amide molecule coordinates with the N-H of a neighboring molecule, forming a hydrogen bond.

	In the very first installment of this column, we learned that infrared peak widths are determined by the strength of the intermolecular interactions between neighboring molecules (6). Hydrogen bonds are an example of a strong intermolecular interaction. As a result, any functional group that engages in hydrogen bonding is going to have broadened peaks compared to non-hydrogen bonded functional groups. We saw this with amines (4), and it applies to amides as well.

	Note that, to the left in Figure 5, the amine N-H stretching peaks come to a point, but are broadened at their base. Similarly, the amide N-H stretching peaks to the right in Figure 5 come to a sharp point, but are broadened at their base. Note that, for both these functional groups, the N-H stretching peaks fall in about the same place, 3400 to 3200 cm

. This means that the N-H stretching peaks by themselves cannot be used to distinguish amides from amines. However, since amides contain a C=O bond and amines do not, these two functional groups are easy to distinguish. N-H stretches with the presence of a C=O stretch indicate amides, N-H stretches without a C=O stretch indicate amines. As we will see in the next column, primary, secondary, and tertiary amides can be distinguished from each other by infrared spectroscopy.

	Amides are an important nitrogen containing functional group. Their structure consists of a nitrogen atom attached to a carbonyl group. There are three types of amide, primary, secondary, and tertiary depending upon the number of C-N bonds present in the group. All amides are conjugated, resulting in their all having a relatively low wavenumber C=O stretching peak. Amides also exhibit hydrogen bonding, resulting in somewhat broadened N-H stretching peaks. The three types of amides can be distinguished from each other by infrared spectroscopy as will be seen in the next column.

https://www.drugabuse.gov/publications/drugfacts/mdma-ecstasymolly

Brian C. Smith, PhD, is founder and CEO of Big Sur Scientific, a maker of portable mid-infrared cannabis analyzers. He has over 30 years experience as an industrial infrared spectroscopist, has published numerous peer reviewed papers, and has written three books on spectroscopy. As a trainer, he has helped thousands of people around the world improve their infrared analyses. In addition to writing for 

, Dr. Smith writes a regular column for its sister publication 

 and sits on its editorial board. He earned his PhD in physical chemistry from Dartmouth College. He can be reached at: 

IR Spectral Interpretation Workshop

Amides are an important functional group found extensively in polymers and proteins. There are three different families of amides. Here, is explained how to distinguish them using infrared spectroscopy.

Organic Nitrogen Compounds VI: Introduction to Amides

X-ray diffraction of crystallized protein continues to be the preferred means of obtaining high-resolution structures of proteins and their complexes. These structures are crucial for drug design, but the screening and optimization of the conditions that produce well-diffracting crystals represent a bottleneck in the structure determination process. In this study, a quantum cascade laser (QCL) infrared microscope was used to determine protein aggregation, distinct from self-association, which is crucial to the success of any crystallization effort. Hyperspectral images of an aliquot from a vapor diffusion hanging drop crystallization screen were acquired over a small temperature range (30–38 ºC), at intervals of 2 ºC. QCL infrared (IR) spectral data were subjected to 2D IR correlation analysis to describe 

 complex (1:1.5 molar ratio) and the selective aggregation of the target peptide. To our knowledge, this is the first time such a level of molecular understanding has been achieved.

	
	Protein crystallization with X-ray diffraction (XRD) is the preferred means of obtaining high-resolution structures of proteins and their complexes. These structures are crucial for drug design, yet methodological barriers to their determination still exist (1−4). Efforts to make the structure determination process more efficient include data collection methods, such as the synchroton beam lines (5), detection via sensitive focal plane arrays, and improved cryogenic and mounting procedures for crystals. However, one key step in successful high-throughput X-ray crystallography continues to be a bottleneck; namely, the screening and optimization of the conditions required to produce well-diffracting crystals. This process includes the optimization of multiple variables, including, but not limited to, precipitant, pH, temperature, protein sequence, and concentration (4). New techniques are needed if we hope to increase the rate of success of this step, which is currently ~18%.

	There are three stages in the crystallization process: nucleation, crystal growth (which is governed by diffusion), and cessation of crystal growth. Crystal growth requires specific interactions between individual protein molecules that lead to an organized crystal lattice. If the crystallizability of the protein of interest is dependent on the type of protein aggregate generated during the crystallization process, then our proposed method should allow for this assessment. Conceptually, the aggregate should be one of self-association, in which chemical associations driven by the weak interactions between proteins or the precipitating agent are critical to maintaining the protein’s conformational stability. Self-association in principle would account for the generation of a crystal lattice, while excluding solvent molecules from the immediate hydration shell of the protein and increasing the influence of the surface charge of the protein to account for the crystallization event. The evaluation of the crystallization process of proteins and protein complexes using this direct and highly informative approach would allow for further understanding of the process of crystallization, and should be explored even further during crystallization optimization as part of the screening process. It may also be possible to evaluate the precipitant’s role in changing the dynamics of the protein’s interaction with its aqueous environment, causing a synergistic effect that will lead to the successful crystallization of the protein or the protein complex.

	Here, we have employed quantum cascade laser infrared microscopy (QCL IRM) for the determination of protein aggregation, distinct from self-association, under various crystallization conditions. The platform includes a QCL infrared (IR) microscope with enhanced signal-to-noise ratio, sampling accessories, cells under thermal control, and software for QCL IR spectral analysis and assessment of crystallizability. Hyperspectral images (HSI) within the mid-IR spectral region of 1800–1000 cm

 provide data distinguishing an aggregate from a crystal, as well as a rough determination of the crystal or aggregate size. The QCL IR spectral data are subjected to 2D IR correlation spectroscopy analysis (6,7) to determine the extent and molecular mechanism of aggregation and  crystallization under thermal stress.

	A high-resolution structure for a yeast homologue of centrin known as CDC31 in the form of CDC31-Sfi1p complex exists as PDB ID: 2DOQ (Figure 1) (8), and our group has extensively studied 

centrin isoforms using multiple biophysical techniques, including differential scanning calorimetry and circular dichroism, Fourier transform infrared (FT-IR), and two-dimensional (2D) IR correlation spectroscopies (9). The current study involved a complex between a centrin variant (

centrin 2 [E32A]) and its target peptide, Sfi1p

Well-diffracting crystals are essential for X-ray diffraction of crystallized protein for structural determination. A quantum cascade laser (QCL) infrared microscope is used to determine protein aggregation, distinct from self-association, for the success of the crystallization effort.

Spectroscopy-11-01-2019

Data Quality and Data Integrity Are the Same, Right? Wrong!

Book Review: Chemometrics in Spectroscopy, 2nd Edition, by Howard Mark and Jerry Workman, Jr.

Vol 34 No 11 Spectroscopy November 2019 Regular Issue PDF

Raman Analysis of Ethylene Vinyl Acetate Copolymers–Using 2D-COS for Identifying Structural Changes

Organic Nitrogen Compounds VI: Introduction to Amides

Quantum Cascade Laser Infrared Microscopy and 2D IR Correlation Spectroscopy Improves Crystallization Screening of a Protein Complex