In the classic article written by Savitzky and Golay (1), which has been cited more than 3800 times according to ISI Web of Science, a kind of digital filter for smoothing and differentiation was developed. In their approach, each successive subset of 2

+ 1 points is fitted by a polynomial of degree 

) differentiation (zeroth differentiation = smoothing) of the original data at the midpoint is obtained by performing the differentiation on the fitted polynomial rather than on the original data. Finally, the running least-squares polynomial fitting can be performed simply and automatically by convolving the entire input data with a digital filter of length 2

The history and development of the Savitzky–Golay (SG) smoothing and differentiation filter have been reviewed in brief (2). The original tabular coefficients of the SG filter have been corrected by Steinier et al. (3) and Madden (4) subsequently. The typo of a formula by Madden (4) has been spotted recently (5). As pointed out by Mark and Workman (5), one limitation is that the existing SG filter is subject to using an odd number of data subset. With a matrix approach, the SG filter was extended recently for even number (2

) data (i.e., the SG filter could have an even length [6]). The feasibility of the even-length SG filter has been validated with computer simulations (6). In the least-square strain estimator (LSQSE) of ultrasound elastography, it was concluded that a data subset of number 8 could obtain the best performance (7). This method is equivalent to the SG differentiation filter with a linear fitting (8), and the data subset of even number corresponds to an even length of the filter. As discussed in (6), for an even order differentiation, the SG filter of even length has a better characteristic of attenuation at the high frequency of ω= π than that of odd length. However, the matrix approach (6) is not very convenient to calculate the filter coefficients. In this article, a more general approach based upon Gram polynomials is developed to calculate the convolution coefficients of the even-length SG filter for all polynomial degrees, all differentiation orders, and all data points.  

) of data subset and assume the index of the data samples ranges from –

. The data samples can be regarded equivalently as being symmetrical about the origin, by shifting the origin one half. Therefore, the input signal are represented by (6) 

  contains the convolution coefficients of the SG filter for different order differentiation at the origin (that is, the imaginary midpoint or the center of symmetry), given by (6) 

 is the fitting polynomial degree, S is the 2

The convolution coefficients for the s-th differentiation evaluated at the imaginary midpoint are given by (6) 

-th column vector of G, the subscript 0 represents the evaluated midpoint. With the matrix approach mentioned earlier, some closed-form solutions as well as the numerical table can be obtained (6). However, only the filter coefficients for the imaginary midpoint are convenient to obtain.  

It is expected that it is difficult to obtain the matrix inversion (S

  in Equation 2 with an explicit formula if 

 is large. Therefore, one would prefer numerical computation to symbolic computation. If 

 S tends to be ill conditioned (9). Hence, a more successful and general approach with the orthogonal polynomials (Gram polynomials or Legendre polynomials) has been proposed (9–11). In this article, such an approach also is extended for an even-length SG filter. All the equations are similar to those in (10), but the effects of the even number data and the aforementioned translation of the origin have been taken into account.  

The smoothing and the differentiation are expressed as 

) differentiation value evaluated at position 

				 are the corresponding coefficients for smoothing and differentiation, respectively. Note that 

 = 0 for the imaginary midpoint of the data subset, while 

– 1/2 for the first, second, and last points, repetitively.  

The convolution coefficients in Equations 6 and 7 are calculated as  

Articles

In various fields such as signal processing, imaging processing, analytical chemistry, and spectroscopic analysis, smoothing and differentiation is important and necessary. With a matrix approach, the Savitzky–Golay smoothing and differentiation filter was extended recently to even length. In this article, a more general approach is proposed for convenient computation.

Savitzky–Golay Smoothing and Differentiation Filter of Even Length: A Gram Polynomial Approach

When you write up your observations and notes in a traditional laboratory notebook, there are a number of features that help to ensure data integrity. The notebook pages are numbered sequentially and they are bound together. Therefore, if a page is removed, it is obvious immediately. When recording observations in the notebook, the order of the write-up is important and is enforced by the sequential or linear pagination of the notebook. The author signs and dates the recorded information and this is witnessed by a second person, who signs and also dates when they reviewed the work. Note that it is not usual to time and date signatures in a laboratory notebook. 

In converting from paper to electronic work, there is not the same linear sequence that occurs in the paper world as the records generated are based around the application software operation. The workflow of the computerized system will define the way that the software works, but users can do many things within this. For example, a data system for a spectrometer can be used to set up the data acquisition parameters, acquire data from the sample, and then process the data to produce results. However, once data have been acquired, a user can review, interpret, and process a spectrum several times before a result is finally accepted. Because of the speed of computers, there is a need to record both the time and the date of any event in the system, typically down to a faction of a second.  

The files and results (both electronic records) produced by a computerized system usually will not be arranged in the linear sequence seen in the laboratory notebook. If you are unlucky, they will be stored in directory structures, which might mean you have to employ file-naming conventions to organize the data. If you are lucky, the system is organized via a database and most of this work is done for you behind the scenes. To capture all creations, modifications, and deletions to data there is an audit trail that acts as the mortar linking all the bricks together to build up the time sequence of events.  

However, fundamental to the operation of the software application and the audit trail is an accurate and precise time and date stamp. Without this it will not be possible to: 					

Link a time and date stamp to any electronic record (for example, file creation, records modification, and so forth)

Establish the integrity of records and actions 

Importance of Accurate Time and Date Stamps

It cannot be emphasized enough that an accurate and controlled time and date stamp is the cornerstone of data integrity within the laboratory and throughout the organization. Without it, you have a situation where you cannot trust anything generated by any computer. ISO Standard 17799 (1) on Information Management Security in section 9.7.3 discusses Clock Synchronization: 

"The correct setting of computer clocks is important to ensure the accuracy of audit logs, which may be required for investigations or as evidence in legal or disciplinary cases. Inaccurate audit logs may hinder such investigations and damage the credibility of such evidence. 

Where a computer or communications device has the capability to operate a real-time clock, it should be set to an agreed standard, e.g. Universal Coordinated Time (UCT) or local standard time.  

As some clocks are known to drift with time, there should be a procedure that checks for and corrects any significant variation." 

This is not a pharmaceutical or healthcare regulation but a general standard for any computerized system operating in any organization. You do this because it is good business sense to do so, not because the regulations say you should. Moreover, in a research environment where spectrometry data supports a patent registration, accurate time and data stamps are important to establish priority in the event of a competing patent application. 

Everyone Believes Computers, Don't They? 

Of course, there is a generally accepted view that when something is performed by computer it must be true simply because the computer says so. In 1990, the U.S. Environmental Protection Agency (EPA) commissioned a report by Hamilton Booz Allen to investigate computer practices by environmental laboratories that reported data to the Agency (2). The use of a third party was significant, as strict anonymity was promised to all contributors — when you see some of the results and working practices in this report, you'll know why. Dubious is one word to describe them; fraudulent is another. As we are concerned with time and date here, we'll focus on just a single feature – time traveling; analysts found that they could travel in time, although it was limited to hours and minutes. 

There are some EPA analyses where time is a critical issue: the sample must be taken and delivered to the laboratory within 6 h. If the sample arrives after this time it must be rejected and a new one taken. However, in a laboratory with a laboratory information system (LIMS) used for sample reception, it is a relatively simple task for a system manager to flip the clock back a few minutes. When the sample was received using the LIMS, though, the time of sample receipt was within the 6-h time window allowed by the EPA. If inspected, the computer says so and must therefore be right.  

The FDA also is interested in time stamps for the same reasons that ISO is: to ensure that any data and electronic records generated by any computerized system such as a spectrometer data system are trustworthy and reliable. Time stamps have been on the periphery of the regulation since the publication of 21 CFR 11 in 1997 (3). Originally, in the preamble to the 1997 regulation, there was a note that time should be local to where the signer was. However, this gave rise to a little problem for systems working across time zones — it was technically impossible. The time stamp for data originates at the server where the data are stored rather than at the client where the data is generated. If the server is located in a different country, you get a different date and time stamp compared to where you are located. 

In a draft Guidance issued in 2001 (4), the FDA changed their stance and stated that systems have to be implemented with a clear understanding of where the time stamp is located. In 2003, in the Part 11 Scope and Application guidance, the FDA stated as a footnote:  

"Although we withdrew the draft guidance on time stamps, our current thinking has not changed in that when using time stamps for systems that span different time zones, we do not expect you to record the signer's local time. When using time stamps, they should be implemented with a clear understanding of the time zone reference used. In such instances, system documentation should explain time zone references as well as zone acronyms or other naming conventions" (5). 

If time and date stamps are so fundamental to data integrity, you might ask, why then is this not a fundamental part of the Part 11 regulation? When the FDA issues the new draft Part 11 — possibly sometime this year or next — perhaps time stamps will be raised from a footnote to an integral part of the regulation.  

The FDA issued a draft guidance for industry on time stamps (4) in March 2002 and then withdrew it in February 2003. Notwithstanding the withdrawal, this guidance has good useful information and the key points and discussions for spectrometer applications are: 					

The time stamp needs to be accurate to "within a minute;" this is best interpreted as ± 1 min.

The location of the time stamp must be defined. For a system working on one site, this is not an issue and is pretty self-evident. However, if you share data between sites in different time zones, then this is critical. Knowing how the system handles times is the key issue here. In the final version of the Part 11 Scope and Application guidance, there is a footnote that confirms that the FDA still wants the location of the time stamp to be defined (5).

The guidance makes no reference to summer and winter daylight savings or to leap years. Typically, these changes will be managed by the operating system, which requires an accurate time and date stamp setting and you might want to check that these have occurred the next working day after the event.  

Because it is fundamental to electronic record integrity regardless of the laboratory you are working in, we need to consider the following issue with time and date stamps: 					

Identification of the time zone and the impact of a data system working across different time zones

How is the time and date checked? This can be a combination of procedural and technical means. In addition, who is authorized to make changes to the date and time, and, as a corollary, who is not allowed to make changes to the time? 

If presented with a date of 05/09/04, what does this mean? It depends where in the world you are and what date format you use. This date could have three meanings: 

1. September 4, 2005 (based upon an ISO standard for date and time that starts with the largest part, the year, and works down) 

2. September 5, 2004 (based upon a European use of date abbreviations) 

3. May 9, 2004 (based upon an American date format) 

An unambiguous date format therefore is required, especially in multinational companies where U.S. and European scientists are working. Personally, I prefer the U.S. military date format (DD MMM YYYY), which means that there is no possibility of confusing the day and month because the month is spelled out, as in 04 SEP 2005. Another alternative is to implement long date formats in the operating system and software application. Here the date format is presented as September 04 2005.  

Format of the time stamp is another issue that should be considered. There are a number of possibilities, and sometimes the seconds can be recorded to 1/100th of a second. Do you use a 12- or 24-h clock to record the hours in the time portion of the stamp? Regardless of your approach, the time and date format must be documented in your user requirements specifications (6). 

 Time stamps for standalone spectrometry systems probably are best set and maintained manually. Although there are technology options, they will be relatively expensive to implement on a number of individual workstations. Therefore, you'll need a manual procedure with records to show that the time stamp has been checked and maintained against a standard time source (for example, a speaking clock or another reliable time source). How often you will need to check the time stamp will vary. As a suggestion, start on a monthly basis for the first six months, then review how accurate the time stamp has been over this period and whether you have needed to adjust the clock. You then can lessen the frequency if necessary on the basis of experience. You will need to maintain a log of who checked the computer clock, the computer time, the standard time, and whether any adjustment was made. This approach can be a waste of time and resources, especially when coupled with the need to manually backup individual systems. Ideally, all standalone systems should be networked to ensure that time stamps are maintained from a single source within the network. 

 This is by far the easier and preferred option for setting and maintaining time stamps. Within your IT network, a server provides the time stamps for all other severs and workstations. At a minimum, each time a user logs on to the network from a workstation, the time stamp is updated automatically. The operating system also can be set up to update the timestamp during the time that a workstation is logged on (for example, hourly, if required). 

Trusted time sources are available to check and correct the network clock automatically; this can occur in a number of ways: 					

Network Time Protocol (NTP), in which the time server accesses a time source on the Internet.

Time sources from the National Observatories of some countries (for example, the U.S. Naval Observatory has atomic clocks in Denver and Washington; there is the Frankfurt atomic clock for Europe; and the U.K. has a source at Rugby).

Columns

Accurate and precise time and date stamps are critical to ensuring integrity of the data and results generated by each computerized system used in any spectroscopy laboratory.

Focus on Quality — Validation of Spectrometry Software: Importance of Accurate Date and Time Stamps

Mass spectrometry (MS) instrumentation has undergone several improvements over the years, including increased sensitivity, ease-of-use, and possibilities for more sophisticated and time-efficient experiments. These developments have led to an increase in the number of biological applications. In the 1990s and until recently, new instrumental innovations in organic MS had focused mainly upon front-end techniques — ionization interfaces specifically. In particular, electrospray ionization (ESI), atmospheric-pressure chemical ionization (APCI), and matrix-assisted laser desorption–ionization (MALDI) have evolved from these developments and have seen significant performance improvements over the years. Along the way, a few other interesting liquid chromatography–mass spectrometry (LC–MS) interfaces (for example, thermospray, particle-beam) have paved the way for modern LC–MS, but have then disappeared quickly from research laboratories. Also, a promising new ionization source, atmospheric-pressure photoionization (APPI) was introduced just a few years ago. As a result of all these innovations, rather good and stable MS sample introduction devices are available for a wide range of compound classes and most researchers seem quite satisfied with what they presently have available at their disposal. Continuing research on these ionization sources currently is aimed at enhancing the performance of such devices, through miniaturization, automation, and parallel-analyses. Lately, however, MS instrumentation research is dominated by improvements in the performance of the actual mass analyzers. Moreover, there is a growing trend toward combining different analyzer designs in order to increase the versatility and allow multiple experiments to be performed simultaneously using a single instrument. One of the most recent examples of such a novel hybrid instrument is the linear ion trap-Fourier transform ion cyclotron resonance mass spectrometer (LIT-FT-ICR), which will be discussed subsequently in more detail. This instrument allows fast, multistage tandem MS (MS-MS) experiments as well as ultra-high resolution and accurate mass determinations. The driving force behind this recent trend of improving analyzer designs mainly has been scientists from biotechnology and pharmaceutical research laboratories. Their time is costly and their analytical protocols now routinely include the use of multiple MS instruments. 

This article illustrates the technical principles and typical applications of the most common mass analyzers used in bioanalytical laboratories today. The classical mass spectrometer — the magnetic sector instrument — is not covered by this tutorial. Although these instruments allow very selective scan modes, their implementation is much less routine and is beyond the scope of this article. Moreover, magnetic sector instruments seem to have reached the apex of their technological development, while improvements in instrument technologies based upon other mass analyzers continue to evolve and develop in new directions. Magnetic sector instruments still are employed for specialized analyses, for example, in the petroleum industry or for the analysis of dioxins by gas chromatography–mass spectrometry (GC–MS). 

There are four basic types of mass analyzers found in modern MS: time-of-flight (TOF), quadrupole and a recent derivative, quadrupole linear ion trap (LIT), quadrupole ion trap (QIT), and FT-ICR. The first part of this article will cover TOF and quadrupole analyzers; Part II will address QIT and FT-ICR. More interestingly, however, there are several very advanced permutations of these analyzers, which through the combination of different analyzers create enhanced MS-MS capabilities, novel scan modes, or superior mass resolution and accuracy. For example, the quadrupole-time-of-flight (QqTOF) instrument has developed into one of the most successful hybrid MS designs, allowing the generation of triple quadrupole–like MS–MS spectra, combined with high resolution and accurate mass capability in the TOF analyzer. This combined potential, together with flexible data-dependent acquisition routines, has been proven to be of significant importance for proteomics MS. Rather than just explaining the underlying physical principles, we will concentrate on their characteristic features, their limitations, and their applications (more details, in particular on the fundamental operating principles, can be found in textbooks on mass spectrometry; for example [1]). In addition, several unique techniques and unusual or exotic mass analyzer designs are highlighted throughout the text. We have chosen all of the examples from our own research; as a result, this overview is biased toward applications to small molecules. Therefore, some interesting recent hybrid designs with mostly proteomics applications, such as ion trap-TOF, only are described briefly. 

A linear TOF mass spectrometer is the simplest mass analyzer that exists today, consisting only of an ion-accelerating region, a flight tube, and a detector. These linear instruments are used primarily in combination with the MALDI technique for analysis of large biomolecules. In theory, all ions experience the same potential difference during acceleration and thus have the same kinetic energy at the start of the flight tube, and thus different velocities depending upon their mass. Therefore, their arrival time at the detector is proportional to their mass and they reach the detector in order of increasing mass. The mass-to-charge ratios (

) of the ions relate to the flight times 

 is the acceleration potential. For example, the electrosprayed antibiotic ciprofloxacin (MW = 331 g/mol; MH+  at 

 332) arrives at the detector 9.3 μs after being accelerated with 20 kV into a 1-m long flight tube, whereas the much heavier singly protonated bovine insulin molecule (

 5734) needs 65.1 Âµs to reach the end of the tube. These flight times translate into the 

 values via Equation 2. The differences of flight times are, of course, the basis for resolving ions of different 

 in the TOF analyzer and the mass resolution depends on the flight time differences, which are proportional to 

The major deficiency of a simple linear TOF instrument is its insufficient mass resolution, resulting from flight time variations of ions of the same 

. First, the laser pulses used for ion generation in MALDI-TOF instruments are fairly long, resulting in a wide distribution of starting times of the same ions. Further, the ionization process adds a certain amount of initial kinetic energy to the molecules before acceleration. Finally, the different spatial positions in the source from where the ions are formed lead to varying values of 

 (Equation 2) and thus to flight time variations. As a result, modern TOF instruments commonly employ two techniques to enhance resolution, 

A reflectron TOF (RETOF) is used to focus ions of the same 

 but of different kinetic energy. There are different designs of reflectrons but often they consist of a series of rings or grids. It is located after the drift tube, creating a retarding field that the ions penetrate. Depending upon their kinetic energy, they enter this field at different depths and then are reflected back into the flight tube, where they drift to the detector, which is placed close to the ion source. Consider the case of ions of the same 

 but significant energy spread, which would be detected as broad peaks in the linear TOF: the ion with a higher kinetic energy penetrates deeper and spends more time in the reflectron. The slow ion, on the other hand, returns to the flight tube faster, but with the same lower kinetic energy. The faster ion will catch up with the slow one at the detector, thereby improving mass resolution greatly. 

The delayed extraction technique can be used on both linear and RETOF instruments, to compensate for the velocity distributions of ions. In this technique, which also is sometimes called 

, ions are allowed to drift free of electric fields after formation for a certain delay time. The ions then are accelerated into the flight tube. By allowing them to drift, faster ions move farther away from the sample target than the slower ions. After the delay, these faster ions then will experience less of the accelerating voltage between the sample target and the extraction plate than the slow ones. This procedure compensates for the initial energy distributions of ions with the same 

There are several benefits that TOF analyzers offer when combined with ionization sources such as electrospray or MALDI. In addition to the high mass range that can be analyzed (with a linear TOF instrument) and the high ion transmission, there is the unique advantage of quasisimultaneous detection of all ions, resulting in high sensitivity "full-scan" analyses (of course, the term "full scan" obviously is wrong in the context of TOF-MS, as a TOF-MS is not a scanning device, but the common terminology is used here for the sake of convenience). Thus, they are ideal in qualitative applications such as the identification of unknown compounds, in which the acquisition of an entire mass spectrum is required. Figure 1 shows an example for the analysis of metabolites of an anticancer drug, analyzed on an ESI-TOF instrument, demonstrating the high resolving power and mass accuracy of a TOF analyzer. In addition, TOF analyzers exhibit very fast acquisition times and thus are ideal for hyphenation with capillary electrophoresis and fast LC. Of course, the pulsed nature of TOF analysis makes the hyphenation with a continuous ion source such as electrospray more difficult than with MALDI, which inherently is compatible with TOF analysis. The necessary "slicing" of the ion beam coming from the ESI source can be performed axially or orthogonally. For organic LC–MS, the design is usually an orthogonal acceleration (oa) TOF type. The ions are focused into the orthogonal accelerator as a narrow ion beam and a slice of it is pushed down into the TOF tube. The amount of ions entering the TOF tube (that is, the length of the "slice") relative to the total amount of ions in the orthogonal accelerator determines the 

 of the instrument (see the sidebar "Duty Cycle in MS" for further explanation). After the heaviest ions have reached the detector, the next pulse pushes another slice of ions into the drift tube. 

 Figure 1. Metabolism of doxorubicin in mouse liver microsomes was analyzed by ESI-TOF for the determination of accurate masses of the metabolites formed in vitro. Idarubicin was used as an internal standard in this experiment and two main metabolites were characterized with the aid of the high resolution afforded by the TOF mass analyzer (with internal calibration).			

The downside of TOF applications is the limited dynamic range; that is, the ratio of the maximum to the minimum observable ion intensities or concentrations over which a linear response is obtained from the detector. This is the direct result of the types of detectors that are employed for handling the large number of high-resolution spectra, making the application of TOF instruments to quantitative analyses less attractive. There is another weakness encountered with TOF analyzers: it is technically difficult to select a precursor ion for MS–MS experiments. In MS–MS, a specific precursor ion is isolated and then made to undergo dissociation to fragment ions, which are analyzed subsequently to obtain structural information. It is important for the precursor ion of interest to be isolated from interfering species; in other words, one needs to be able to select a very narrow 

 range. In a RETOF instrument, a TOF gate can provide only a very coarse precursor selection (over several 

 units wide), to study metastable compositions. This technique is used in combination with postsource decay (PSD) and is quite popular for sequencing of peptides. 

It is much more elegant, however, to utilize a separate mass spectrometer for precursor ion selection, allowing the isolation of a single 

 value. One option is to use another TOF drift tube together with a time selector, and such a design is available commercially for specialized proteomics applications (see the sidebar "TOF–TOF Instruments for High Energy CID"). Much more common, however, is the implementation of a quadrupole mass filter prior to the TOF tube. In fact, this can be taken one step further, by adding a second, radio frequency (rf)-only quadrupole as a collision cell (Q

 TOF). The TOF analyzer then is used to provide accurate mass data for fragment ions formed in q

 . Before we describe the QqTOF instrument, however, we will take a look at the basic operating principles of a quadrupole. 

TOFâTOF Instruments for High-Energy CID			

The quadrupole is the most common analyzer found in analytical laboratories today. It is not only the standard analyzer for GC–MS (which will not be discussed here) but also is part of various designs of LC–MS instruments. Quadrupoles are not used only as mass analyzers; they also are implemented frequently as ion transfer optics, collision cells, and linear ion traps (

). This mass analyzer works on the basis of electric fields generated between a set of four axial rods through which ions pass on their way to the detector. The voltages applied to these rods consist of direct current (dc) and rf components, which together create a quadrupolar field, allowing only a certain 

 range to pass through for a given combination of potentials; ions outside that 

 region hit the rods and are discharged. The quadrupole therefore is referred to also as a 

, into which ions are accelerated from the ion source by a small potential. Positive ions will be attracted to negative rods, and vice versa. The rf component, however, causes the electric field to alternate at a frequency in the megahertz range. The applied potential to the rods is of the form: 

 is the rf potential of frequency ω/2π. Two rf waveforms are applied to two pairs of opposing parallel rods, with a 180° phase shift. As a result, the trajectory of an ion fluctuates constantly as it travels between the rods until its detection.  

In a full-scan experiment, the dc and rf components are ramped at a constant ratio, and ions entering from the ion source are enabled to pass through the rod assembly successively. For a specific combination of dc and rf, only a very small 

 range can pass through, the size of which determines the resolution of the instrument. Quadrupoles usually are operated at unit resolution throughout the mass range (often up to 

 1500 and 1501, for example. Higher mass resolution, however, is available on a recent commercial quadrupole instrument produced by Thermo Electron (Waltham, MA). The sequential detection in the full-scan mode of any quadrupole analyzer results in a low duty cycle, typically of the order of 0.1%, depending upon the monitored 

 range. The duty cycle (and the resulting sensitivity) is much higher in the 

 (SIM) mode of the quadrupole (close to 100%). In this case, the dc and rf potentials are held constant, so only a specific 

 ratio can pass through. One also can jump between different voltages, to allow more than one 

 ratio to be detected sequentially, with a corresponding reduction in duty cycle. 

Single quadrupole instruments usually are limited to measuring intact species generated by the ionization source, resulting in limited selectivity in comparison to many of the advanced designs described in the following. They have, nevertheless, found a wide application range in LC–MS and GC–MS. Such instruments often are considered more of a selective HPLC detector today than a mass spectrometer. Typical applications of single quadrupoles include purity assessments, molecular weight confirmations, and automated preparative HPLC fractionations. Figure 2 shows an example for a mass-triggered fractionation experiment from a semipreparative HPLC run, where a single quadrupole was used to supply trigger signals to the fraction collector, to start collecting the chromatographic peaks of interest. 

Figure 2. Mass-triggered fractionation chromatogram using a single quadrupole instrument set-up with specific m/z values for triggering the fraction collector. A crude phytoplankton extract was purified into several fractions for further analysis of marine toxins by mass spectrometry. Arrows above show the start of individual triggers (with the m/z responsible for the trigger).			

In a single-quadrupole instrument operating with an ESI or APCI source, selectivity can be increased when fragmentation is induced in the ion source region ("in-source CID"). This procedure refers to the activation of ions in the region between the ion source and the analyzer, in which fragmentation can be initiated by collisions with residual gas molecules at intermediate pressures. The generated fragment ions can be used for limited structure elucidation or confirmation purposes. One must be careful, however, as no precursor ion selection takes place and fragment ions cannot be linked specifically to the precursor ion of interest. In other words, fragment ions could originate from several compounds for coeluted peaks in LC–MS, making structure assignments difficult (similarly, quantitative analysis in the SIM mode would be difficult when isobaric interferences contribute to the chromatographic peaks). 

The most common mass analyzer for quantitative bioanalytical analyses is the 

. This MS–MS instrument consists of three sequential quadrupoles (Q

 ; Q refers to a mass-resolving quadrupole, q to an rf-only quadrupole; 

). This configuration allows additional ion activation in q

 , after the ion of interest has been selected in Q

  is operated in the rf-only mode, in which only an rf voltage is applied to the quadrupole rods (no dc component), thus effectively becoming a wide-band pass for the ions. It can be filled with a neutral gas such as N

  or Ar, acting as a collision gas. The ions leaving Q

  with offset voltages between 0 and 100 V, typically. The results of the 

 (CID) can be analyzed with the mass analyzer Q

 . This is the simplest form of an MS–MS experiment on a triple-quadrupole mass spectrometer, and it is called a 

. In addition, several other selective MS–MS modes are available, depending upon which mode Q

 essentially is the reverse of the product ion scan. The third quadrupole is set to select a specific product ion formed in q

 , and the first quadrupole is scanned for all precursor ions forming the chosen fragment. An important use for precursor ion scans is seen in the pharmaceutical industry, for the identification of drug metabolites with similar fragmentation behavior to that of the parent molecule. A 

 experiment also is achieved in this instrument with relative ease. Neutral loss scans are used routinely to identify common functional groups present in a set of molecules. Figure 3 shows one of the most common applications conducted with triple-quadrupole instruments, the quantitative analysis of drugs and their metabolites in biological specimens such as plasma or urine. In these analyses, the 

 is almost always applied, allowing enhanced sensitivity and selectivity by circumventing isobaric interferences from the biological material. The experiment shown in this example actually was run on a QqLIT instrument, where the third quadrupole is replaced by a linear ion trap (see below). For MRM analyses, however, both these instruments function in the same manner. 

Figure 3. Representative chromatogram from a human serum sample after administration of morphine. The following MRM transitions were monitored for quantitation of the parent drug and its two glucuronide metabolites: morphine (m/z 286â201), morphine-d3 (internal standard, m/z 289â201), morphine-3-glucuronide and morphine-6-glucuronide (m/z 462â286), and morphine-3-glucuronide-d3 (internal standard, m/z 465â289).			

Interestingly, in addition to their function as collision cells, rf-only quadrupoles also are widely used as ion transfer optics. In that application, ions passing through the quadrupole are confined in the center of the quadrupole axis at low gas pressures, thus increasing transmission efficiency. Often, other multipoles such as hexapoles or octopoles are used for the same purpose. This effect is called 

. Furthermore, by adding trapping electrodes at the beginning and the end of such a quadrupole, one can store ions in the device. These 

 have become very popular recently, either as a stand-alone MS or as front-end stages prior to FT-ICR, or as Q

  in a triple-quadrupole MS. Among several other scan modes, the QqLIT design allows an additional stage of MS-MS after the regular triple-quadrupole CID. Figure 4 shows an illustrative example of how a QqLIT can add complementary information to the regular CID analysis in q

 . There is, however, a catch to these experiments: trapping ions in a LIT takes quite a bit of time, in comparison to the residence time of ions in a regular QqQ instrument. For the QqLIT instrument, that would result in a significantly reduced duty cycle when additional MS

  experiments are performed in the LIT, because it requires some time for the resonance excitation to take place (typically 50–200 ms). During that time, ions coming from q

  would be lost. If we trap the ions in the meantime, however, and transfer the ion bundle into Q

  are finished, we would not lose any ions, and duty cycle would be restored. In the commercial QqLIT instrument, this is achieved by storing the ions temporarily in a focusing quadrupole (before the QqQ; the manufacturer calls it q

Figure 4. The MS3 spectrum of doxorubicin above was performed using a QqLIT instrument. Secondary fragments of the product ion at m/z 397 are shown in the lower trace and give information on the fragmentation pathways of this anticancer drug. Note that the ion at m/z 398 is due to the isolation of the M+1 isotope peak of m/z 397 in the LIT, without subsequent excitation by on-resonance CID, therefore it is still in the linear ion trap when the MS product ions are detected.			

  with a completely different mass analyzer, in particular, with a high-resolution device. This makes a lot of sense, because it allows the acquisition of accurate mass data for fragment ions. The most common hybrid system is the quadrupole-quadrupole-time-of-flight (QqTOF) MS. The coupling of the Qq section and the TOF system is achieved in exactly the same way as described in the previous section for ESI-TOF, that is, by adding an orthogonal acceleration region after the collision cell. The TOF section now serves only to record spectra, with high resolution and high mass accuracy. QqTOF systems were developed originally for peptide sequencing applications, but also are now applied routinely for the accurate mass determinations of small molecules; for example, for confirmation of tentatively identified metabolites during the drug discovery process. An example for the QqTOF characterization of fragment ions formed from CID analysis of a protonated biotoxin molecule is shown in Figure 5. 

Figure 5. Product ion spectrum of neosaxitoxin on a QqTOF instrument. The experimental m/z values for several product ions are shown above with their corresponding mass accuracies, measured in parts per million (ppm) (note that the mass accuracy is the difference between the measured accurate mass and the calculated exact mass. It is often expressed in relative terms, ppm).			

3. R.N. Zare, F.M. Fernandez, and J.R. Kimmel, 

This tutorial illustrates the technical principles and typical applications of the most common mass analyzers used in bioanalytical laboratories today, as well as novel techniques and mass analyzer designs. Examples are based upon the authors' research in small molecule applications.

Tutorial — Mass Analyzers: An Overview of Several Designs and Their Applications, Part I

Sensors based upon laser absorption spectroscopy enable sensitive and selective detection of gas phase species in a wide range of applications (1). Concentrations in the parts-per-million (ppm) to parts-per-trillion (ppt) ranges (2) can be detected, using relatively compact laser-based devices. Applications include detection of trace moisture in semiconductor process gases, monitoring of atmospheric pollutants with significance in the occurrence of disease, breath analysis for diagnosis of asthma, and measurement of hydrocarbons for combustion diagnostics. A variety of techniques can be used to implement the required detection capability. For fence-line monitoring applications, such as for emission monitoring at chemical plants, long-path absorption measurements can provide concentration integrated over distances of hundreds of meters — in this case a laser beam is directed at a retroreflector and the return signal is measured by a detector adjacent to the source. For in situ sensing, where measurement of the species is performed in proximity to the optical source, prominent absorption techniques are photoacoustic detection, tunable diode laser absorption spectroscopy (TDLAS) using a multipass cell, and cavity ring-down spectroscopy (CRDS). 

All of the techniques mentioned here share a common critical requirement for a tunable, single-frequency laser operating in continuous wave (CW) mode. The laser frequency must be tuned to the peak of the absorption feature of the species of interest. Furthermore, its spectral bandwidth must be a small fraction of the width of the absorption feature. These requirements are met most commonly using semiconductor lasers. Using different semiconductor gain media, external cavity diode lasers (ECDLs) using a tunable grating to provide feedback can generate single-frequency output at wavelengths between 630 nm and 2 μm. In addition to ECDLs, single-frequency output also can be obtained, with limited tunability, from monolithic devices such as distributed Bragg reflector (DBR) or distributed feedback (DFB) diodes. These devices exhibit compactness, ruggedness, and high reliability, partly derived from their development for telecommunications applications. 

However, for ultrasensitive trace gas detection, it is desirable to operate laser-based sensors at mid-infrared (IR) wavelengths above 2 μm. This is because many gas species have their fundamental vibrational transitions here, in the so-called "molecular fingerprint region." Although gas sensors can be developed successfully using near-infrared sources, the detection sensitivities of such sensors are much lower than those based in the mid-IR, because the absorption line strengths of the fundamental transitions are orders of magnitude higher than those of the corresponding near-IR overtone and vibrational combination transitions (see Figure 1). 

 Figure 1. Water vapor absorption strength increases from near to mid-infrared (HITRAN data). Vibrational band assignments for the upper state are shown for several features.			

Hence, the development of practical mid-IR CW sources is something of a "holy grail" for the sensing community. Much effort and resources have been poured into pushing the spectral coverage of semiconductors into this region. Lead-salt lasers (3) have for years been the only technology available, with wavelength coverage from 3–30 μm. However, their low power output and requirement for cryogenic cooling have largely kept these devices from emerging from the research laboratory. Greater promise of practical mid-IR sources for sensor applications is shown by interband cascade lasers (1) (operating between 2 and 5 μm), and quantum cascade (QC) lasers (4) (above 4 μm). Both types of devices have shown many of the characteristics needed, including high power and narrowband operation. However, to date, although there have been demonstrations of room-temperature CW operation of QC lasers at certain wavelengths, these devices have yet to reach a mature stage of commercial development. As a result, there remains a substantial current need for laser sources that enable basic spectroscopy and sensor development in the mid-IR. 

The most promising alternative to semiconductor sources is in nonlinear frequency conversion of near-IR tunable lasers. There are two principal methods of mid-IR generation through the use of nonlinear materials. These are difference frequency generation (DFG) and optical parametric oscillation (see Figure 2). Both techniques allow exchange of energy between three optical frequencies, known as pump, signal, and idler (here 

 ). The interacting waves abide by the energy conservation relation, shown in Figure 2. Difference frequency generation is a process of frequency mixing two single-frequency sources (for example, ECDLs) on a single pass through a nonlinear material to generate tunable narrowband output at a third frequency defined by the difference in energy between the two input frequencies. This technique has been used successfully as the basis of mid-IR sensors in studies of atmospheric chemistry (5). The principal limitation of the technique is that the efficiency of conversion in the CW regime is extremely low, with maximum output power levels measured typically in microwatts (μW) to milliwatts (mW). 

Figure 2. Difference frequency generation (DFG) and optical parametric oscillation (OPO) allow frequency conversion from near-IR to mid-IR. 			

By contrast, optical parametric oscillators (OPOs) can provide highly efficient conversion of a near-IR pump laser to longer wavelength output, while retaining the spectral and spatial properties of the pump laser. OPOs long have been regarded as promising tunable coherent sources, since their initial demonstration in the 1960s. However, their development in the commercial arena has been slow, with viable products emerging only in the late 1980s. Even then, OPOs have been available only as pulsed devices, mainly in the nanosecond pulse regime, with output typically extending from ~400 nm to 2500 nm. Although single-frequency, pulsed OPOs are available commercially, these devices are large, expensive systems occupying a substantial fraction of an optical table, and their linewidths are substantially broader than those of CW systems. Hence, beyond applications such as LIDAR, which requires high pulse energy output, they are not well suited to incorporation into practical sensors. 

In the continuous wave regime, progress has been slower (6). Although there have been recent attempts to launch CW OPO products, the limited tuning capability of the devices offered has not allowed them to tap successfully into the spectroscopic market. Progress in development of CW devices has been held back by limitations in available pump sources, nonlinear materials, and in the OPO configurations, which could be implemented using them. Ideally, a CW OPO pump source is required to possess essentially all the characteristics that are required of a spectroscopic source in the target OPO output wavelength region — that is, single-frequency spectral output, high beam quality, continuous tunability, and high power. Although various sources can meet some of these demands, until recently, none have been able to provide all of these characteristics. Simultaneously, it is only with the recent development of periodically poled nonlinear materials that crystals now provide enough gain to allow CW oscillation using practical pump lasers. Early demonstrations of CW operation of OPOs were performed with unwieldy sources such as argon ion lasers, and required that the OPO resonated both the signal and idler waves within the OPO cavity. Although demonstration of stable operation and tuning of such devices has been performed (7), the requirement introduces unwelcome complication to the device. However, using periodically poled materials including lithium niobate, operation of singly resonant devices has been enabled (8, 9). These devices can be tuned straightforwardly, either by varying the poling period or temperature of the nonlinear material, or by tuning its pump source. This latter method is particularly attractive, in that it removes the tuning mechanism entirely from the OPO, and takes advantage of well-developed frequency tuning schemes that have been demonstrated in mature near-IR pump sources. Hence, it is with the advances in development of semiconductor lasers and fiber amplifiers, made under the auspices of the telecom industry, that the final leap has been made to CW OPOs fully capable of meeting spectroscopic application needs. 

Fiber amplifiers and lasers have made spectacular advances within a period of a few years to the point of being able to generate kilowatt-level CW power in a diffraction-limited beam, while single frequency sources such as DFB semiconductor and fiber lasers provide high spectral quality and tunability, as well as high reliability. 

Figure 3. A fiber-pumped CW OPO consists of rack-mounted fiber pump laser and frequency converter head.			

Combining these advances, a CW OPO has been developed (Figure 3) pumped by an all-fiber pump source, which demonstrates the characteristics demanded of high-resolution, mid-IR spectroscopy. The pump source for the OPO is a single-frequency fiber laser, which can be tuned by applying piezoelectric strain to the fiber. The power from this source is amplified to a level of several watts by a fiber amplifier. Both sources are configured as packaged, rack-mounted units, similar to those developed for telecom applications. This provides a pump source in which the light is confined entirely within the fiber, without any possibility of misalignment, which can be a problem in typical diode-pumped, solid-state lasers. The OPO is configured as a ring-cavity, singly resonant OPO, with the shorter, signal wavelength resonant. This signal wavelength is held fixed by an intracavity etalon, so that any tuning of the pump source is directly reproduced by the OPO idler wavelength. Hence, rapid mode-hop-free tuning capability with a range of over 50 GHz is available in the spectral output of the mid-IR OPO idler. Similarly, the high spatial and spectral quality of the pump also is transferred to the OPO idler, with diffraction-limited output and linewidth below 1 MHz. Up to 85% of the 1-μm photons are converted to 3-μm photons within the OPO, so power levels of up to 2 W have been measured using only 6 W of pump power (Figure 4). 

Figure 4. 2.7-Î¼m OPO power output versus fiber laser power input.			

Cavity ring-down spectroscopy (CRDS) is an ideal technology platform for the application of the fiber-pumped OPO. The high sensitivity of the CRDS technique is achieved by coupling laser light into a high finesse Fabry-Perot cavity containing a gas sample (Figure 5). Long effective path-lengths are obtained within the cavity, and absorption is measured by monitoring the time taken for light to leak out of the cavity when the laser is switched off. In the presence of an absorbing species, the "ring-down" time is shortened. The development of the fiber-pumped CW OPO is sponsored by the Optoelectronic Manufacturing group at the National Institute of Standards and Technology (Boulder, CO), as a means to monitor trace impurities in process gases that affect semiconductor manufacture yield. Kris Bertness of NIST has demonstrated a measurement sensitivity of 50 nmol/mol in CRDS measurement of residual moisture in bulk phosphine gas (10). This measurement was made with a near-IR semiconductor laser probing relatively weak combination band transitions of water vapor at 938 nm. The motivation for the development of an OPO operating in the 2.7 μm region has been to allow the same technique to be used while accessing absorption features, which are as much as 100 times stronger (see Figure 1). This potentially can result in a similar increase in detection sensitivity. Such capability can allow rapid, sensitive monitoring of trace water to provide a sensitive measure of real-time variation. This capability should lead ultimately to improvements in semiconductor source gas purity and hence, improved manufacture yield. 

Figure 5. Cavity ring-down technique measures the rate of leak-out of light trapped in an optical resonator, containing a gas sample. 			

Further characteristics of the OPO (Figure 6) make it particularly suited to the cavity ring-down technique. First, power levels are very high relative to diode lasers used typically in the technique — power is measured in 100s of milliwatts to watts. Second, the excellent beam quality enabled by the fiber pump source allows much better mode-matching of the laser into an optical cavity. Better mode matching allows more photons to be coupled into the fundamental transverse mode of the ring-down cavity. This improved coupling efficiency results in higher signal throughput and improved absorption signal-to-noise ratio (S/N), ultimately translating into better measurement precision. The detection sensitivity of the CRDS technique also can be enhanced by the narrow linewidth of the source laser. Because the spectral width of the ring-down cavity resonances is extremely narrow, the fraction of incident photons coupled through the cavity is increased as the linewidth of the laser approaches that of the ring-down cavity resonance. As a result, again the S/N of the detected absorption signal is increased. CW OPO linewidths typically are measured in kilohertz, while typical diode laser linewidths are in the megahertz range. 

Figure 6. Fiber-pumped CW OPO provides narrow linewidth, wide tunability, and diffraction-limited beam quality.			

Third, the broad tuning range permits the selection of strong impurity absorption lines in regions of the lowest interference from the host gas, and increases the likelihood that the same system can be used to monitor multiple impurities. In monitoring moisture in phosphine, for example, the ultimate sensitivity limits are set by the ability to reliably account for the phosphine off-resonant absorption line background. The frequency-stabilized cavity ring-down apparatus used at NIST Boulder was developed originally by colleagues at NIST Gaithersburg (11). Unlike conventional implementations of CRDS, this approach uses the stabilized frequency comb of the ring-down cavity to provide frequency markers, thus enabling low-uncertainty determination of line intensities, and high-resolution measurements of line shapes, as well as measurements of pressure shifting and Doppler-free saturation effects. Project leader Joseph T. Hodges also has worked with the specialty gas industry to perform characterization of the performance of a commercial near-IR-based ring-down system (12) in similar moisture concentration measurements (13). According to Hodges, "tunable single-frequency mid-IR CW laser sources are critical to the realization of high-sensitivity absorption spectroscopy of residual species in gas mixtures. Coupling such spectroscopic-grade light sources with cavity-enhanced and high-spectral resolution absorption methods like CRDS, can enable quantitative gas sensing of trace analytes by discriminating weak analyte absorption from background absorption interferences associated with bulk gases."  

Under the NIST development program, initial demonstrations of the OPO in spectroscopic measurement, including the use of cavity-enhanced techniques, have been performed. Collaborative experiments have been performed with Professor Houston Miller of the department of Chemistry at George Washington University (GWU), Washington, DC. The Miller group at GWU has applied cavity enhanced absorption techniques to the detection of trace constituents of air as well as to the development of combustion diagnostics. For applications in this field, the operation of the OPO in the 3–4 μm band is advantageous due to the existence of the fundamental "C–H stretch" vibration in this wavelength region, which allows very sensitive detection of many hydrocarbons and combustion products. In preliminary experiments performed at Aculight (Bothell, WA), the tuning capability of the OPO was illustrated (Figure 7) by generating single-pass spectra of carbon dioxide at 2810 nm (~3559 cm

 ). Demonstration of efficient coupling into a ring-down cavity also was performed at the latter wavelength. By scanning the laser frequency over a free spectral range of the ring-down cavity, it was demonstrated that the linewidth of the OPO at 3167 nm was less than 1 MHz (Figure 6). Based upon the sensitivity of previous near-IR ring-down measurements at GWU, we anticipate potential detection sensitivities in the sub-part-per-billion regime using the high absorption strengths at this wavelength. This continuing collaborative work is expected to realize the potential for versatile, ultrasensitive, multi-species trace gas detection for a wide variety of applications. 

Figure 7. Single-pass absorption spectra for 60-cm pathlength of CO2 (5%, 50 torr) and CH4 (5%, 30 torr) measured using fiber-pumped OPO. Symbols: experimental data; lines: fits to HITRAN 2004. 			

Future Development of CW OPOs for Spectroscopy

It is anticipated that future directions in the development of this technology will include expansion of the wavelength coverage to wavelengths as long as 10 μm using novel nonlinear materials such as optically patterned gallium arsenide (OP-GaAs). OPOs using this material already have demonstrated wavelengths throughout this range using pulsed pumping. Development of more sophisticated electronic controls will provide capabilities for frequency locking, stabilization, and automated tuning, allowing application of the devices to a wider range of spectroscopic techniques. Finally, further investment producing reduction in cost might be the most important factor, enabling the use of OPOs in a wider range of real-world sensor applications beyond initial adoption as a laboratory spectroscopy tool. 

, I. Sorokina and K. Vodopyanov, Eds. (Springer-Verlag, Berlin, 2003). 

2. G. von Basum, D. Halmer, P. Hering, M. Murtz, S. Schiller, F. Muller, A. Popp, and F. Kuhnemann, 

5. D. Richter, A. Fried, B. Wert, J. Walega, and F. Tittel, 

 "Slow progress for OPOs in tunable systems market," March 2002. 

7. A. Henderson and R. Stafford, "CW optical parametric oscillator pumped by a distributed feedback fiber laser" postdeadline paper CPDA7 in Conference on Lasers and Electro-Optics 2004. 

8. I. Lindsay, B. Adhimoolam, P. Gross, M. Klein, and K.-J. Boller, 

9. A. Henderson and R. Stafford, "1.2 W Threshold, Singly Resonant Fiber-Pumped CW OPO: A High-Power Mid-Infrared Spectroscopic Source" paper ALT-4, Solid State and Diode laser technology review, Directed Energy Professional Society (2005). 

10. K. Bertness, S. Lehmann, and J. Hodges, 

13. V. Vorsa, S. Dheandhanoo, S. Ketkar, and J.T. Hodges, 

 is development engineer, at Aculight Corporation (Bothell, WA). 

 is with the Department of Chemistry, George Washington University (Washington DC). 

New mid-infrared spectroscopic sources, based upon advances in fiber lasers and in nonlinear frequency conversion, are now enabling high-resolution laser spectroscopy in the 2–4 μm wavelength region and beyond. With this in mind, the authors discuss continuous wave (CW) optical parametric oscillators (OPOs) in particular.

Spectroscopy-11-01-2005

Savitzky–Golay Smoothing and Differentiation Filter of Even Length: A Gram Polynomial Approach

Focus on Quality — Validation of Spectrometry Software: Importance of Accurate Date and Time Stamps

Tutorial — Mass Analyzers: An Overview of Several Designs and Their Applications, Part I

Continuous Wave Optical Parametric Oscillators Break New Spectral Ground