The Raman spectra of a particular face of a single crystal can be significantly different if acquired with different microscope objectives. The purpose of this installment of “Molecular Spectroscopy Workbench” is to inform and educate users of micro-Raman instrumentation of the effect of the microscope objective on the Raman spectra of crystals. Furthermore, we explain the underlying physics of changes in relative intensity and even peak position of certain Raman bands depending on the microscope objective used to acquire the spectrum. Changes in peak position are attributed to phonon directional dispersion sampled through wide-angle microscope objectives with different numerical apertures.

	Laser illumination and light collection of Raman spectrometers can involve mirrors, simple lenses, fiber-optic probes, and microscope objectives. The foremost consideration pertaining to these optical elements is how effectively they deliver the laser illumination and collect the Raman scattered light, and how strong the Raman signal is for a given laser power density and spectral acquisition time. Regarding the light-collection optics in particular, the important characteristics include the solid angle of collection and optical throughput. A spectroscopist’s expectation is that the overall signal strength may vary depending on the collection optic used but that the Raman spectrum-that is, the relative intensities of the bands–will not. That expectation is certainly reasonable for Raman spectra acquired from liquids, gases, amorphous materials, glasses, and polycrystalline solids. However, that expectation is not always valid when performing Raman spectroscopy of single crystals or grains. One can observe variations of the peak positions and the relative intensities of Raman bands depending on the optic used to deliver the laser illumination and collect the Raman light. This phenomenon is most evident in the Raman spectra acquired with different microscope objectives. The purpose of this installment of “Molecular Spectroscopy Workbench” is to inform and educate users of micro-Raman instrumentation about the effect of the microscope objective on the Raman spectra of crystals.

	One can find detailed theoretical and experimental treatments on the effect of microscope objectives and wide-field collection optics that were published in the early days of micro-Raman spectroscopy (1–3). An excellent source of information on this topic is the chapter written by George Turrell titled “Raman Sampling” in the book 

 published in 1988 (4). Turrell’s very instructive chapter deals with laser excitation focusing and Raman light collection using wide-angle microscope objectives. Of course, one generally thinks of the microscope objective’s magnification as one of its most important and relevant characteristics. One should also be mindful of the objective’s numerical aperture (NA) and its direct relationship to Raman light collection efficiency. The numerical aperture is defined by the following expression:

 is the refractive index of the medium through which the light is passing and θ is the angle between the ray along the optical axis at the center of the lens and the ray from the perimeter of the lens to the sample. Therefore, a microscope objective with a high NA will have a greater solid angle of light collection. However, remember that the NA does not just affect the light collection efficiency. One can envision a variation of the solid angle of the laser beam incident on the sample depending upon the NA of the objective. At the very highest magnification and NA, the laser beam comes to a focus at a very short working distance over a wide angle. This means that the laser beam passing through a 100x objective with a 0.9 NA focused on the sample will not have the same illumination profile as that from a 25-mm-diameter lens with a 40-mm focal length. The microscope objective profile can be thought of as a wide-angle cone consisting of varying angles of incidence, whereas the long-focal-length lens more closely approximates a collimated narrow beam of parallel rays orthogonal to the sample surface. Furthermore, if we define the direction of illumination as the 

-axis, one can expect that the laser light polarization initially in the 

-plane will now have some component along the 

-axis at the wider angles of incidence of a microscope objective. The effect of the microscope objective’s NA on illumination and light collection has significant consequences for the practice of micro-Raman spectroscopy as applied to crystals. We demonstrate these effects through spectra of single-crystal LiNbO

 acquired from objectives with different magnifications and numerical apertures.

	Raman Spectroscopy of Single-Crystal LiNbO

	In performing Raman spectroscopy of single crystals, it is important to understand the nomenclature of crystallography. The three crystallographic axes are generally labeled 

 by optical physicists. The crystallographic face is defined by the crystallographic axis normal to that face; that is, the face and axis are orthogonal to each other. For example, one is looking at the 

-face of a crystal if the line of eyesight is parallel to the 

-axis. In a cubic crystal, all three axes have the same refractive index. A uniaxial crystal has two axes with the same refractive index, called the 

, has a different refractive index from the ordinary axes. All three of the crystallographic axes of a biaxial crystal have different refractive indices. Uniaxial and biaxial single crystals are ideal for demonstrating the consequences of using wide angle microscope objectives on crystalline materials. LiNbO

 is a uniaxial crystal whose ordinary and extraordinary refractive indices at 530 nm are 

 = 2.2355, respectively (5). A correlation exists between the directionally dependent dielectric properties and the corresponding Raman tensors of single crystals. Moreover, the Raman spectrum acquired from a single crystal depends on the crystal class to which the crystal belongs, the crystal’s orientation relative to the direction and polarization of the incident light, and the collection angle and polarization of the Raman scattered light.

 crystal class for which the Raman tensors are

	where the letter in parentheses next to the symmetry species indicates the crystallographic direction of the lattice vibrational mode polarization. The irreducible representations of the LiNbO

 modes are infrared and Raman active, whereas the 5

 modes are silent. Raman spectra acquired from the 

-face of a single crystal of uniaxial LiNbO

 are shown in Figure 1. Spectroscopic selection rules predict that these Raman bands will belong to the symmetry species 

(TO), where the symbols TO and LO stand for the transverse and longitudinal optical modes, respectively. The spectra were acquired using 10x (NA 0.25), 50x LWD (NA 0.50), 50x (NA 0.75), 100x LWD (NA 0.75), and 100x (NA 0.90) microscope objectives. The laser beam was incident upon the 

-axis, one of the ordinary axes. No Raman analyzer was used in the collection of these or any other spectra presented in this work. The five spectra are plotted normalized and appear to be nearly identical, independent of the microscope objective used for laser delivery and light collection. This is the type of response, independent of microscope objective, which many users of Raman instrumentation may incorrectly expect of all crystallographic Raman measurements. However, rotate the incident polarization by 90° such that the incident laser polarization is now parallel to the 

-axis, the extraordinary axis, and one observes entirely different Raman spectra. The 

 mode now dominates the spectra shown in Figure 2.

-axis polarized spectra are significantly different from those obtained with the incident light polarized parallel to the 

-axis. Most of the bands appearing in the 

-polarized spectra of Figure 1 are attributed to the 

 symmetry species, whereas the most pronounced bands in the 

-polarized spectra of Figure 2 belong to the 

 symmetry species. These differences arise because of the Raman polarization selection rules and the interplay of incident and scattered light direction, polarization, and the orientation of the crystal. Selection rules predict that these 

-face Raman bands excited with laser light polarized along the 

-axis in the backscattering configuration will belong to the symmetry species 

(TO). Most of the Raman bands appear to be of near identical relative strength, independent of the microscope objective used to acquire the spectrum. However, notice that the intensity of a single band at 195 cm

 depends on the microscope objective used to deliver the laser beam and collect the Raman scattered light. The 195 cm

 band is absent in the spectrum acquired using the 10x objective and increases progressively with the 50x LWD, 50x, 100x LWD, and 100x objectives. Moreover, the 195 cm

(LO) symmetry species and is not expected in this configuration (6–9). Why would a Raman band forbidden in this experimental configuration appear in the spectra and why would its relative intensity vary depending on the microscope objective used? The answer to this question can be found in Table I, which shows the numerical apertures of the objectives used.

 is a measure of the solid angle of light collection of the microscope objective. A large NA value corresponds to a greater solid angle. In this case, the NA has implications for the laser beam profile incident on the sample. Ideally, one would like to fill the back aperture with the laser beam such that it comes to a diffraction limited spot in the field of view. Not all the rays envisioned travelling down the barrel of the microscope objective will reach the laser spot at the same angle of incidence. The objective with the greatest NA will cause the rays passing through the lens near the perimeter to refract at the largest angle. Therefore, one can envision a distribution of angles of incidence on the sample that is dependent on the numerical aperture. The field of view may be normal to the 

-face, but that does not mean that all of the rays from the objective will be parallel to the 

-axis. Some will be incident on the sample at wide angles commensurate with the numerical aperture. Thus, we see with increasing NA the emergence of the 195 cm

(LO) symmetry species and should not be observed under this 180° backscattering experimental configuration. These types of differences in Raman spectra that are dependent on the objective used can easily go unnoticed by one not skilled in the practice of Raman spectroscopy. The purpose of this installment is to make users aware of this phenomenon and to explain its origin.

	Rotation of the crystal by 90° about the 

-axis allows the laser beam to now be parallel to the extraordinary axis (

-axes are the ordinary axes with identical refractive indices. Spectra were acquired using 10x (NA 0.25), 50x LWD (NA 0.50), 50x (NA 0.75), 100x LWD (NA 0.75), and 100x (NA 0.90) microscope objectives and are shown in Figure 3. Selection rules predict that these 

(TO). The spectra have been normalized to the 

, thereby revealing the changes in relative intensities of all the other bands as a function of the microscope objective used. The contrast with the spectra in Figure 1 is quite stark where there were no differences among the spectra acquired with different objectives. Here, we see that all of the Raman bands, to a greater or lesser degree, vary with intensity depending on the objective used. All of the Raman bands in Figure 3 follow the same pattern. The strength of a given band increases relative to that of the 872 cm

 band progressively with the 10x, 50x LWD, 50x, 100x LWD, and 100x objectives. It is important to note that all the bands do not increase in signal strength by the same proportion relative to the 872 cm

 band. For example, the bands at 155 and 582 cm

 band in the spectrum acquired with the 10x objective. However, in the spectrum acquired using the 100x objective the signal strength of the 155 cm

 band now significantly exceeds that of the 872 cm

 band whereas the signal strength of the 582 cm

 band is still less than that of the 872 cm

 band. There are other ways of viewing and analyzing this same phenomenon; we could have plotted the spectra normalized to any of the other bands to see the relative signal strength variations dependent on the microscope objective. The variations in relative signal strengths are because of a change in the distribution of angles of incidence as we use objectives with increasing numerical aperture. The 10x objective with NA of 0.25 comes the closest of those used here to deliver a laser beam parallel to the crystallographic axis. As the NA of the objective is increased, one can envision more rays incident on the sample at oblique angles rather than just those closely parallel to the crystallographic axis. Consequently, Raman bands that would be expected were the laser beam to be incident at an oblique angle with respect to the light-collection axis appear unexpectedly in the backscattering configuration.

	Not only do we observe a change in the relative intensities, some bands manifest a shift in peak position as an objective with higher numerical aperture is used. The spectra of Figure 3 are shown in an expanded scale in Figure 4 to more clearly see the relative intensity changes and, in some cases, peak shifts. Specifically, note that the 153 and 578 cm

 bands in the spectrum acquired using the 10x objective shift to 155 and 582 cm

, respectively in the spectrum acquired with the 100x objective. The bands at 237, 264, 332, and 432 cm

 are fixed, showing no shift when using different microscope objectives. What accounts for the shift in peak position of some bands but not others? The answer is 

	Consider the case of laser light propagating along a general direction in a crystal and not parallel to a crystallographic axis. Such light can excite lattice vibrational waves that resolve into longitudinal and transverse waves. The frequency of the longitudinal phonon is higher than that of the transverse phonon. If our phonon is propagating in a general direction, we may expect to see a splitting and two Raman bands associated with the LO and TO phonons. That behavior can be expected if the plane contains only the two ordinary axes of the uniaxial crystal. However, the situation is quite different if the exciting radiation is propagating in a plane containing the extraordinary axis, either the 

-planes. Now the short-range atomic forces that lead to the extraordinary refractive index and different vibrational force constants of the unique axis play a role. If these forces are greater than the long-range forces that lead to the splitting of longitudinal and transverse phonons, then a phonon of mixed longitudinal and transverse character will be launched by the exciting radiation. Thus, instead of a splitting of the Raman bands because of the LO and TO phonons, one will observe a single Raman band of mixed LO and TO character dependent on the direction of propagation in the crystal. Imagine a laser beam initially directed along the 

. If Raman spectra were acquired as the crystal is rotated about the 

-axis in small rotational increments until the laser beam was finally directed along the 

-axis, one would observe that some peaks shift to higher wavenumber as the laser illumination rotates from the 

-axis. Those lattice vibrational modes most affected by short range atomic forces will be of mixed LO and TO character and their energies will depend on the direction of propagation through the crystal lattice. This phenomenon is known as 

. A detailed theoretical treatment can be found in the book by Sherwood (10).

	A plot of experimentally obtained phonon directional dispersion of the Raman bands of LiNbO

 is shown in the publication by Maimounatou and colleagues (9). We see the greatest dispersion for the 

 when the exciting beam is parallel to the 

(TO) modes, respectively as the crystal is rotated 90° such that the incident beam is finally parallel to the 

-axis. The phonon directional dispersion shown in Figure 4 of the Maimounatou publication is consistent with the change in Raman peak position as microscope objectives with higher numerical aperture and greater off-axis illumination is used.

	Raman Spectroscopy of Single-Crystal KTiOPO

 is a biaxial crystal whose three refractive indices at 530 nm are 

 is a ferroelectric crystal at room temperature with high nonlinear susceptibilities and electrooptic coefficients. Both crystals are commonly used for frequency doubling. The 532-nm laser is produced by guiding the 1064-nm beam from a Nd:YAG laser through a crystal such as LiNbO

 oriented appropriately to achieve phase matching. Here we continue our examination of the effect of microscope objective numerical aperture on the Raman spectra of KTiOPO

	The irreducible representations of the vibrational modes are

 modes are Raman active, leading to a Raman spectrum much more complicated than that of LiNbO

. Nevertheless, we see the same phenomena and patterns as a function of microscope objective as we did with our spectra of LiNbO

-face with the laser beam guided along the 

-axis with the light polarized parallel to the 

-axis are shown in Figure 5. The spectra were acquired using 10x (NA 0.25), 50x LWD (NA 0.50), 50x (NA 0.75), 100x LWD (NA 0.75), and 100x (NA 0.90) microscope objectives. All of the spectra are plotted normalized to the 765 cm

 band. It is clear that many of the bands manifest the same intensity relative to that of the 765 cm

 band independent of the microscope objective used. However, note how the strengths of the bands at 268, 327, 637, and 702 cm

 vary depending on the objective used. In particular, the 702 cm

 band is absent in the spectrum acquired with the 10x objective. It appears first in the spectrum acquired using the 50x LWD objective and grows progressively more intense as the numerical aperture increases through the 50x, 100x LWD, and 100x objectives. The behavior is much like that of the LiNbO

	Expansion of the scale in the spectral region below 425 cm

 shown in Figure 6 allows one to more clearly see the changes in relative strength of the bands at 268 and 327 cm

 when compared to those invariant at 342, 375, and 400 cm

. The latter are of identical strength independent of the microscope objective used to acquire the spectrum. We observe the same progression in these spectra of KTiOPO

, depending on the microscope objective used. The changes in relative band strength can be accounted for and explained with respect to the increasing numerical aperture. As the numerical aperture increases, so does the distribution of angles of incidence of the exciting laser beam.

	We observed the phenomenon of phonon directional dispersion in our spectra of the 

 and so we might expect to observe the same effect in our spectra of KTiOPO

, especially with the exciting beam nominally parallel to the 

 = 1.8873 and is substantially greater than 

 = 1.7924. Therefore, the short range and anisotropic atomic forces leading to this birefringence should lead us to expect phonon directional dispersion in KTiOPO

. Indeed, one can see in Figure 6 that the band at 328 cm

 obtained with the 50x LWD objective shifts to 327 cm

 in the spectrum acquired using the 100x objective. An expanded view of these same spectra in the region from 500 to 850 cm

 is shown in Figure 7. The band at 768 cm

(LO) mode (12). Here we see that the 517, 547, 701, and 834 cm

 bands remain fixed, independent of the objective used to acquire the spectra. However, the intense band at 768 cm

 acquired using the 10x objective shifts progressively to a lower wavenumber with increasing numerical aperture until reaching a value of 765 cm

 in the spectrum acquired with the 100x objective. The progression of phonon directional dispersion to a lower wavenumber with increasing numerical aperture observed for KTiOPO

 is the opposite of that observed for LiNbO

 where the bands shifted to a higher wavenumber. We can attribute the different responses to the different anisotropic short-range atomic forces of KTiOPO

-axis or extraordinary refractive index (

 is less than that of the ordinary refractive index (

 = 1.7924. Consequently, the phonon energies shift in a direction commensurate with the positive or negative values of the birefringence.

	The polarization of the incident beam also plays an important role. The spectra of the 

-face shown in Figures 5–7 were all obtained with the incident laser light polarized parallel to the crystallographic 

-face with incident laser light polarized parallel to the 

-axis are shown in Figure 8. The spectra appear similar but are not identical to those shown in Figure 7. In particular, note that the phonon directional dispersion of the 

(LO) band is greater as it progressively shifts from 768 cm

 acquired using the 10x objective to lower wavenumber with increasing numerical aperture until reaching a value of 763 cm

 in the spectrum acquired with the 100x objective. The phonon directional dispersion of the 

(LO) mode is even greater for light polarized parallel to the 

-axis than it is for light polarized along the 

-axis. Therefore, one needs to be cognizant of the polarization of the incident laser light along with the numerical aperture of the microscope objective when acquiring Raman spectra with a microscope based instrument.

	The numerical aperture is a measure of the solid angle of light collection of the microscope objective. A microscope objective with a high NA will have a greater solid angle of light collection. The NA also has implications for the laser beam profile incident on the sample. The objective with the greatest NA will cause the rays passing through the lens near the perimeter to refract at the largest angle. Therefore, one can envision a distribution of angles of incidence that is dependent on the numerical aperture. Consequently, Raman bands that would be expected were the laser beam to be incident at an oblique angle with respect to the light collection axis appear unexpectedly in the backscattering configuration. We observe a change in the relative intensities and some bands manifest a shift in peak position as an objective with higher numerical aperture is used. This phenomenon can be attributed to phonon directional dispersion. The field of view may be normal to a particular crystal face, but that does not mean that all of the rays from the objective will be parallel to that crystallographic axis. Therefore, one needs to be cognizant of the numerical aperture of the microscope objective when acquiring Raman spectra of single crystals or grains with a microscope based instrument.

		C. Bremard, J. Laureyns, J.-C. Merlin, and G. Turrell, 

		C. Bremard, P. Dhamelincourt, J. Laureyns, and G. Turrell, 

, D.J. Gardiner and P.R. Graves, Eds. (Springer-Verlag, Berlin, 1989), pp. 13–54.

Physics and Chemistry of Crystalline Lithium Niobate 

		S. Margueron, A. Bartasyte, A.M. Glazer, E. Simon, J. Hlinka, and I. Gregora, 

		S. Sanna, S. Neufeld, M. Rüsing, G. Berth, A. Zrenner, and W.G. Schmidt, 

		B. Maimounatou, B. Mohamadou, and R. Erasmus, 

 (Cambridge University Press, London, 1972), pp. 83–115.

Ferroelectric Crystals for Laser Radiation Control

		M. Rüsing, C. Eigner, P. Mackwitz, G. Berth, C. Silberhorn, and A. Zrenner, 

 is a Raman applications manager at Horiba Scientific, in Edison, New Jersey, where he works with Fran Adar. David is sharing authorship of this column with Fran. He can be reached at: 

Articles

Molecular Spectroscopy Workbench

The Raman spectra of a particular face of a single crystal can be significantly different if acquired with different microscope objectives. This article explains the underlying physics of changes in relative intensity and even peak position of certain Raman bands depending on the microscope objective used to acquire the spectrum.

The Effect of Microscope Objectives on the Raman Spectra of Crystals

general chapter <1058> “Analytical Instrument Qualification” has been updated and became effective August 1, 2017. So, what has changed in the new version?

	In the regulated world of good manufacturing practice (GMP) we have regulations that define what should be done, but leave the interpretation to the individual organization on how to do it. However, when we come to the regulated analytical or quality control (QC) laboratory we also have the pharmacopoeias, such as the 

 to provide further information to help interpret the regulations. These tomes can have monographs for active pharmaceutical ingredients, finished products, and general chapters that provide requirements for how to apply various analytical techniques, such as spectroscopy.

has a general chapter on analytical instrument qualification (AIQ) (1). This chapter came about with a 2003 conference organized by the American Association of Pharmaceutical Scientists (AAPS) on analytical instrument validation. The first decision of the conference was that the name was wrong and it should be analytical instrument qualification (AIQ). The conference resulted in a white paper (2) that after review and revision became 

general chapter <1058> on AIQ effective in 2008. This chapter described a data quality triangle, general principles of instrument qualification, and a general risk classification of analytical equipment, instruments, and systems. General chapter <1058> did not specify any operating parameters or acceptance limits because those can be found in the specific general chapters for analytical techniques.  
	
	This column is written so that you will understand the changes that come with the new version and the impact that they will have on the way that you qualify and validate instruments and the associated software, respectively.

	Why Is Instrument Qualification Important?

	The simplest answer to this question is that qualification is important so that you know that the instrument is functioning correctly and that you can trust the results it produces when it is used to analyze samples.  
	
	However, there is a more important reason in today’s world of data integrity–integrated instrument qualification and computer validation is an essential component of a data integrity model. The complete four-layer model can be viewed in my recent book, 

Validation of Chromatography Data Systems 

(3), but the analytical portion of the model is shown in Figure 1 and is described in more detail in an earlier “Focus on Quality” column (4).

		Foundation: management leadership, policies and procedures, culture, and ethos

		Right instrument and system for the job: instrument qualification and computer validation

		Right analytical method for the job: development and validation of analytical procedures

		Right analysis for the right reportable result: analysis from sampling to reporting the result 

	
	The model works from the foundation up with each layer providing input to the next. As shown in Figure 1, AIQ and computerized system validation (CSV) come after the foundation layer, illustrating that if the instrument is not qualified and any software is not validated, the two layers above (method validation and sample analysis) will not be effective and can compromise data integrity. Interestingly, AIQ and CSV are missing from many of the data integrity guidance documents but you can see the importance in the overall data integrity framework of Figure 1.

<1058> had several issues that I presented during a session co-organized by Paul Smith at an AAPS conference in 2010 and published in this column (5). There were three main problems with the first version:

		Problem 1: The true role of the supplier was missing. The supplier is responsible for the instrument specification, detailed design, and manufacture of the instrument but this responsibility is not mentioned in <1058>. The reason is that the section on design qualification (DQ) mentions that a user can use the supplier’s specification. However, a user needs to understand the conditions under which the specification was measured and how relevant it is to a laboratory’s use of an instrument.

		Problem 2: Users are responsible for DQ. 

<1058> places great emphasis on the fact that in the design qualification stage is the responsibility of the supplier, “Design qualification (DQ) is most suitably performed by the instrument developer or manufacturer.” This is wrong. Only users can define their instrument needs and must do so to define their intended use of the instrument and comply with GMP regulations (§211.63) (6).

		Problem 3: Poor software validation guidance. The software qualification and validation description in 

<1058> is the poorest part of this general chapter as software is pervasive throughout Group B instruments and Group C systems. 

	Although the approach to handling embedded software in Group B instruments where the firmware is implicitly or indirectly validated during the instrument qualification is fine, there are omissions. Users need to be aware that both calculations and user defined programs must be verified to comply with GMP requirements in 211.68(b) (6). Note that the qualification of firmware, which is a simple and practical approach, is now inconsistent with Good Automated Manufacturing Practice (GAMP) 5, which has dropped Category 2 (firmware). 

	Software for Group C systems is the weakest area in the whole chapter <1058>. The responsibility for software validation is dumped on the supplier: “The manufacturer should perform DQ, validate this software, and provide users with a summary of validation. At the user site, holistic qualification, which involves the entire instrument and software system, is more efficient than modular validation of the software alone.”

	In the days of data integrity, this approach is completely untenable. The United States Food and Drug Administration (FDA) guidance on software validation (7), quoted by <1058>, was written for medical device software, which is not configured unlike much of the laboratory software used today.

	To try and rectify some of these issues, the revision process of 

<1058> started in 2012 with the publication of a “Stimulus to the Revision” process article published in 

 written Chris Burgess and myself (8). This article proposed two items:

		An integrated approach to analytical instrument qualification and computerized system validation (AIQ-CSV), and

		More granularity for Group B instruments and Group C computerized systems to ensure that calculations and user defined programs were captured in the first group and an appropriate amount of validation was performed by the users for the second group.

	We used the feedback from that article to draft a new version of 

 <1058> in the summer of 2013, which was circulated to a few individuals in industry and suppliers for review before submission to the 

	Proposed drafts of the new version were published for public comment in 

in 2015 (9) and 2016 (10) and comments were incorporated in the updated versions. The approved 

<1058> final version was published in January 2017 in the 

 40 (11). The chapter became effective on August 1, 2017. There was an erratum published in February, but the only change was reference of the operational qualification (OQ) testing to the intended use definition (user requirements specification [URS]).

	First let us look at the overall scope of changes between the old and new versions of 

	The following items were omitted from the new version of <1058>:

		Differences between qualification and validation. This was omitted because qualification and validation activities are integrated in the new version, so why describe the differences? You need to control the instrument and any software and if you can demonstrate this through the 4Qs process described in the new <1058>, why bother with what the activity is called? 

		Table I in the old version of <1058> describes the timing, applicability, and activities of each phase of AIQ and it has been dropped from the new version. Rather than give a fixed and rigid approach to AIQ, as the table did, there is more flexibility in the new version of <1058> and omitting this table reinforces the new approach.

		Standalone software: Because standalone software has nothing to do with AIQ, not surprisingly this section has been omitted from the new version. The GAMP 5 (12) and the accompanying good practice guide for laboratory computerized systems (13) or my chromatography data system (CDS) validation book (3) will be adequate for this task.

		Examples of instruments in the three categories: Because the instrument classification depends on the intended use there is no need to give a long list of instruments or systems in Groups A, B, and C. It is the intended use of the instrument that defines the group and providing a list produces anomalies. For example, in the old version of <1058> a dissolution bath is listed in Group C when it should be in Group B if there is only firmware and the instrument is calibrated for use. To avoid these arguments the list of examples has been dropped from the new version.

	Of greater interest to readers will be the changes and additions to the new general chapter, again these can be seen in Table I. Below I will discuss the following three areas that reflect the main changes to the general chapter:

		Changes to DQ, installation qualification (IQ), and OQ phases and how this impacts your approaches to AIQ

<1058> update to the “Roles and Responsibilities” section makes users ultimately responsible for specifying their needs, ensuring that a selected instrument meets them and that data quality and integrity are maintained (11). The manufacturer section now includes suppliers, service agents, and consultants to reflect the real world of instrument qualification. One new responsibility is for the supplier or manufacturer to develop meaningful specifications for users to compare with their needs. Incumbent on both users and suppliers is the need to understand and state, respectively, the conditions under which specifications are measured to ensure that laboratory requirements can be met. We will discuss this further under the 4Qs model in the next section.

	Finally, there is a requirement for a technical agreement between users and suppliers for the support and maintenance of any Group B instrument and Group C system. The agreement may take the form of a contract that both parties need to understand the contents of and the responsibilities of each.

 <1058> uses the same 4Qs model as the 2008 version. Yes . . . but there are some significant differences. Look at Figure 2, which presents the 4Qs model in the form of a V model rather that a linear flow. This figure was also published in a “Questions of Quality” column authored by Paul Smith and myself (14). However, Figure 2 has now been updated to reflect the changes in the new version of 

 <1058>. Look at the green-shaded boxes to see the main changes:

	Design qualification now has two phases associated with it. 

		The first phase is for the users to define the intended use of the instrument in a user requirements specification: Users must define functional and operation specifications and intended use (11). Although the new <1058> notes that this definition is expected to be minimal for commercially available instruments it does not mean slavishly copying supplier specifications-especially if you do not know how any of the parameters have been measured. The output from this process is a URS or your intended use definition.

		The second phase is the qualification of the instrument design. This means that you confirm that the selected instrument meets your design specification or intended use. If looking outside of the analytical laboratory, medical device manufacturers call this process 

-ensuring that the users’ current requirements are in the system and any omissions are mitigated, where appropriate.

	These two sections are where most laboratories get it wrong for reasons such as we know what we want (therefore why bother to document it) or we believed the supplier’s literature. This is where most qualifications fail because there is no specification upon which to base the testing in the OQ phase of the process as shown in Figure 2. Executing an OQ without a corresponding URS or design document is planning to fail any qualification. This is one of the major changes in the new version of USP <1058>.

		Note that when the new <1058> talks about minimal specifications for commercial instruments, it does not include minimal specifications for software used to control them for Group C systems. Here you not only need to consider the control of the instrument, but also the acquisition, transformation, storage, and reporting of data and results that includes how data integrity and quality is assured. This activity is not expected to be minimal-especially in today’s regulatory environment.  

		There is also the possibility, for Group B instruments, of merging the laboratory user requirements and the DQ: the decision to purchase being made on the URS (with the proviso that the contents are adequate).

		Risk management is implicit in the <1058> classification of the instrument groups and the subgroups of B and C instruments and systems but more needs to be done in the specification and configuration of the software. For example, access controls, data acquisition, and transformation are key areas for managing data integrity risks.

		What is also shown in Figure 2 is that if there is any change of use during the operation of the instrument or system, it must trigger a review of the current specifications with an update of them, if appropriate.  

	In the new version of <1058>, installation qualification now includes

		The installation of software and the involvement of the IT function to interface an instrument to a network 

		The requirement for conducting an IQ for nonqualified instruments is replaced with a requirement to gather available information and conduct a risk assessment to determine if an IQ should be conducted. In many cases, if an instrument has been installed and maintained by a supplier with records of these activities, but has not been formally qualified, the risk assessment may determine that no IQ should be performed, placing increased emphasis on the OQ phase to demonstrate fitness for intended use. 

		The introduction to <1058> mentions that activities can be merged-for example, IQ and OQ (11). I would see suppliers taking advantage of this option to have a single IQ and OQ document for ease of working, because both phases are typically conducted by the same individual. The combined protocol must be preapproved by the laboratory and then post-execution reviewed by them. However, the post-execution review needs to be conducted while the engineer is still on site so that any corrections can be done before the individual leaves the site.

	Operational qualification has also been extended to include

		A new section on software functions and the differences between software configuration and customization for Group C systems.   

		It is important to configure software (and document the settings) before an OQ is conducted, otherwise you’ll be repeating some tests. In practice, however, there may be a differentiation of duties because a supplier may only perform a basic qualification of the unconfigured software leaving the laboratory to configure the application and then conduct further verification of the whole system. This point is critical because data integrity gaps are usually closed through configuration of the application. However, unless the application is known or a copy is already installed and configured, it is unlikely that the OQ will be performed on a configured version because the laboratory may not know the process to be automated (for example, hybrid operation, electronic operation, or incorporate spreadsheet calculations into the software).

	The major changes to this General Chapter occur in the section on software validation. They are shown diagrammatically in Figure 2. Because the instrument examples have been removed in the new version of <1058> and replaced with the need to determine the group based on intended use, a formal risk assessment now needs to be performed and documented. A risk assessment, based on Figure 3 to classify instruments based on their intended use has been published by Burgess and myself (15) and is based on the updated classification used in the new version of 

<1058> (11). As can be seen from Figure 2, the risk assessment should be conducted at the start of the process in the DQ phase of work because the outcome of the risk assessment can influence the extent of work in the OQ phase.

Focus on Quality

The new version of United States Pharmacopeia general chapter  “Analytical Instrument Qualification” became effective August 1, 2017. What does this mean for you?

Spectroscopy-09-01-2017

The Effect of Microscope Objectives on the Raman Spectra of Crystals

What’s New in the New USP ?

The Carbonyl Group, Part I: Introduction

The 2017 Emerging Leader in Molecular Spectroscopy Award

In Situ Raman Spectroscopy Monitoring of the Reaction of Sulfur Trioxide with Polyethylene Fibers in Chlorinated Solvents

Vol 32 No 9 Spectroscopy September 2017 Regular Issue PDF