Optimizing UV laser focus profiles for improved MALDI performance



Matrix assisted laser desorption/ionization (MALDI) applications, such as proteomics, genomics, clinical profiling and MALDI imaging, have created a growing demand for faster instrumentation. Since the commonly used nitrogen lasers have throughput and life span limitations, diode-pumped solid-state lasers are an alternative. Unfortunately this type of laser shows clear performance limitations in MALDI in terms of sensitivity, resolution and ease of use, for applications such as thin-layer sample preparations, acceptance of various matrices (e.g. DHB for glycopeptides) and MALDI imaging. While it is obvious that the MALDI process has some dependence on the characteristics of the laser used, it is unclear which features are the most critical in determining laser performance for MALDI. In this paper we show, for the first time, that a spatially structured laser beam profile in lieu of a Gaussian profile is of striking importance. This result enabled us to design diode-pumped Nd : YAG lasers that on various critical applications perform as well for MALDI as the nitrogen lasers and in some respects even better. The modulation of the beam profile appears to be a new parameter for optimizing the MALDI process. In addition, the results trigger new questions directing us to a better understanding of the MALDI process. Copyright © 2006 John Wiley & Sons, Ltd.


In the past 20 years, mass spectrometry has become an important tool in biochemistry with the invention of electrospray ionization (ESI)1 and matrix assisted laser desorption/ionization (MALDI).2–4 MALDI, as it is used typically today with solid organic matrices, was invented by Hillenkamp and Karas5 in 1988. Since then MALDI has become an important desorption and ionization technique for large, nonvolatile molecules.6, 7 Originally MALDI was used with axial time-of-flight (TOF) instruments in linear and reflector mode and later extended to orthogonal TOF, Fourier transform ion cyclotron resonance (FTICR) and ion trap technologies.8 Since then great effort has gone into describing the desorption and ionization processes characteristic of the MALDI technique.9, 10

While the original MALDI development work utilized quadrupled Nd : YAG lasers at 266 nm wavelength, most analytical work employed nitrogen lasers at 337 nm. The reason for this is based in economics. Nitrogen lasers were much less expensive and smaller than the Nd : YAG lasers of the day and so all major manufacturers of MALDI-TOF instruments designed their systems based on nitrogen lasers. In parallel, several groups tried alternative lasers and wavelengths to study the MALDI process.11, 12 Even infrared light was used successfully.13 But since commercial systems were built around nitrogen lasers, a whole generation of users optimized their research and applications on the special characteristics of the nitrogen laser.

More recent applications in fields of proteomics and genomics14, 15 including MALDI imaging16–18 and LC-MALDI have created the demand for high throughput instrumentation. Although the nitrogen laser has been analytically very successful over the years, this type of lasers inherently has two significant limitations for high-throughput applications: the maximum repetition rate of current commercial lasers is limited to about 50 Hz, and the average life span of a nitrogen laser is only 2 × 107–6 × 107 shots. In contrast, solid-state lasers such as diode-pumped Nd : YAG lasers can easily achieve rates > 1000 Hz and have a life span of typically several 109 shots. Because high-throughput applications such as MALDI imaging and LC-MALDI demand higher acquisition rates and longer laser life spans, diode-pumped solid-state lasers, especially frequency-tripled Nd : YAG lasers (355 nm), being smaller and more economically attractive than their flash-lamp-based predecessors, offer an alternative to nitrogen lasers.


The Nd : YAG laser has favorable properties, but its performance is clearly inferior to that of the nitrogen laser in some applications. For example, Nd : YAG lasers have been successfully employed for MALDI analysis of peptides using α-cyano-4-hydroxycinnamic acid (HCCA) dried-droplet preparations, but the analytical performance for other important applications such as sinapinic acid (SA) preparations of proteins and 3-hydroxypicolinic acid (3-HPA) preparations of oligonucleotides (unpublished) is generally not as good as that observed using a nitrogen laser. Additionally, while Nd : YAG lasers work well with HCCA dried-droplet preparations, the results are significantly worse using HCCA thin-layer preparations when compared with a nitrogen laser.19 Clearly these differences in performance limit the utility of solid-state lasers in many of the more demanding applications. It is therefore crucial that we understand the properties of the nitrogen laser that make it a better choice in certain cases. We have configured a MALDI-TOF instrument with nitrogen and Nd : YAG lasers to investigate these differences and devise methods for making their MALDI performance comparable. In conducting this study we have looked beyond the obvious wavelength differences and found that the beam profile provides many of the advantages of nitrogen lasers for many applications. In characterizing the involved phenomena we have measured the beam profile at the target surface and examined the MALDI performance for three different cases: (1) the nitrogen laser at 337 nm, ‘N2’; (2) the frequency-tripled Nd : YAG laser at 355 nm with a Gaussian beam profile, ‘Nd : YAG (Gaussian)’; and (3) the frequency-tripled Nd : YAG laser, at 355 nm, using a structured focus profile, ‘Nd : YAG (structured A) and Nd : YAG (structured B)’. This ‘smart’ modulation of the beam profile used in case (3) has been given the name ‘smartbeam’ by Bruker. Such a modulation of the beam profile provides an important additional parameter for defining MALDI setups and, as we demonstrate, can produce dramatic improvements to the MALDI performance of solid-state lasers.



Acetonitrile (ACN) and acetone were purchased from Sigma-Aldrich and trifluoroacetic acid (TFA) from Fluka. The peptide calibration standard, an oligonucleotide calibration standard (PAI Model heterozygous 4G/5G), α-cyano-4-hydroxycinnamic acid (HCCA) and 3-hydroxypicolinic acid (3-HPA) were obtained from Bruker Daltonics. Ion-exchange resins were obtained from Bio-Rad.

Sample preparation

Thin-layer preparation of peptides

A saturated solution of HCCA in acetone was prepared and applied to the stainless steel target resulting in a thin layer of the matrix. The peptide calibration standard was dissolved in 0.1% TFA/water and mixed 1 : 1 with HCCA saturated in 30% ACN/0.1% TFA/water. One microliter of this solution was applied to the matrix thin layer resulting in the deposition of 2 pmol peptide mixture on the target, which was dried at ambient air.

Preparation of oligonucleotides

The kit content containing three oligonucleotides between 5.9 and 6.6 kDa was dissolved in 50 µl of water. 3-HPA was dissolved in water (50 g/l) and ammonium-loaded ion-exchange beads were added to remove the sodium ions. A 1 : 20 dilution of the sample was mixed 1 : 1 with the matrix solution, and 1 µl was applied to a stainless steel target and dried at ambient air.

Preparation of proteins

A saturated solution of SA in ethanol was prepared and applied to a stainless steel target resulting in a thin layer of the matrix. A 1 : 1 mixture of the protein solutions (BSA and RNAse B) and SA in 30% ACN/0.1% TFA in water was applied to the thin layer of SA and dried in ambient air.

MALDI mass spectrometer

A Bruker Ultraflex II MALDI-TOF mass spectrometer with a modified laser setup was used, which allowed switching among N2, Nd : YAG (Gaussian) and Nd : YAG (structured A or B) lasers within seconds. All laser beams were directed into the instrument using the same beam path, thus hitting the sample in the ion source at exactly the same angle (30°) and position. This was a prerequisite for the direct comparison of the behavior of the different laser types on the same sample in the same instrument. For Nd : YAG (Gaussian) and Nd : YAG (structured A or B) the same diode-pumped and frequency-tripled Nd : YAG laser was used (Azura, MALDI-200; max. pulse energy 300 µJ; beam diameter 0.4 mm; divergence 1.4 mrad; TEM00,); only the devices for modulating the beam profile were added or removed. For the nitrogen laser we used an LTB MNL 205 M/C system (max. energy 150 µJ; beam dimension 1 × 2 mm2; beam divergence 4 × 5 mrad).

Generation of structured beam profiles

Figure 1 shows the general setup of the optics for modulating the beam profile. The laser beam first passes through the attenuator (A) to adjust the laser fluence to the intended value, followed by the modulator (M). The modulator shapes the profile using interference principles. In order to mimic the N2 laser, which we have found to present a different profile for each laser pulse, the modulator is designed to vary the position of the structures from shot to shot, by moving the whole structure a few micrometers sideways. Lens 1 (L1) focuses the patterns to an image plane. Here the outer interference structures are cut off with an aperture, to limit the boundaries of the profile. In this setup, the exact final shape of the profile critically depends on details of the modulator, the exact position of L1 and the exact axial adjustment of the entire laser optics. From the variety of experiments we conducted, we know that having a structured beam as such is important, but the exact structure is not so important for obtaining the results. In this paper, we therefore have focused most effort in measuring and describing the profile shape used in the different experiments, rather than describing how the different parameters affected the shape. The image plane is then imaged onto the target with the help of lens 2 (L 2) and at the same time reduced by a factor of 8 in size. To change from Nd : YAG (structured A or B) to Nd : YAG (Gaussian), the modulator and the aperture were removed from the laser beam. In this case, L1 was used to set the desired diameter of the Gaussian profile. To change from Nd : YAG to N2, light from the nitrogen laser was deflected coaxially with the beam of the Nd : YAG, as shown in Fig. 1. For using N2, the modulator and aperture were removed from the beam and L1 was positioned to provide far-field focusing on the target.

Figure 1.

Optical setup for modulation of the laser profile (intensity distribution of the laser beam on the target surface) and for direct switching among the Nd : YAG (Gaussian), Nd : YAG (structured A or B) and N2. Legend: A = attenuator; M = modulator; L1 = lens 1, f = 50 mm; O = round aperture (600 µm diameter); L2 = Lens 2, f = 50 mm; T = MALDI sample target, L2′ = lens 2′, f = 50 mm; L3 = lens 3, f = 50 mm. All lenses are spherical. The elements needed only for measuring of laser profiles are in green (drawing not to scale). Planar mirrors, which are needed for precise axial beam adjustments, have been omitted for simplicity.

Setup for measuring the beam profile and laser power

For measuring the shape of the beam profile at the plane of the target surface, a portion of the beam is coupled out and guided through the same optical setup as the beam traveling into the source, as shown in Fig. 1 in green. Lens L2′ is at the same distance from the aperture as lens L2. Instead of placing the CCD camera in the virtual target plane, this plane is magnified by a factor of 4 and then focused on the CCD camera using L3 for high-resolution visualization of structures on the same scale as the 5-µm CCD pixels. Additionally, this parallel setup allowed recording of the profiles simultaneously with the measurements.

For calibration of this setup, we replaced the standard circular aperture with an aperture in the shape of the letter ‘E’. The ‘E’ then was illuminated with a wide Gaussian beam of the Nd : YAG laser. After proper adjustment, a well-defined ‘E’ pattern was burned in the matrix on the target. At the same time, a well-resolved ‘E’ shape was generated on the CCD camera. This proved that the beam profiles observed in both image planes were simultaneously generated and were identical on the target and on the CCD camera (Fig. 2). This also allowed the calibration of the reproduction scale. The size of the burn marks on the target was measured by the movement of the xy-stage by an appropriate distance. The absolute size of the ‘E’ on the target was determined to be 80 µm × 45 µm, providing the calibration reference for the determination of the FWHM diameter of a laser focus on the target. The mere inspection of the burn marks on the target would be difficult, as the ‘half maximum’ typically is poorly defined.

Figure 2.

Calibration and check of the optical setup: (a) 80 × 45 µm burn pattern of an ‘E’ on the target, inset: 3× magnification; (b) the same ‘E’, simultaneously recorded on the CCD chip for measuring focus profiles. The colored scale references the different relative intensities and is valid for all color coded figures in this paper.

The CCD camera produced bitmaps from single laser shots. We used home-made software based on Labview (National Instruments) to read out the camera picture and to record and display these bitmaps. Saturation of the images was avoided by image acquisition control in the software. The resulting bitmaps were displayed in several views as shown in the ‘Results’ section.

Direct determination of the laser energy deposited on the target was performed by opening the ion source and placing an energy meter (type PEM 100 from LTB, Berlin) in lieu of the target. This was repeated for the different lasers, all used attenuator settings and positions of Lens L1, with and without the modulator and the aperture. Therefore absolute laser energies were obtained from the experiments with errors of less than 10%.


Thin-layer preparation is an important technique for proteomics applications,20 as it results in high sensitivity and allows for in situ sample purifications that greatly facilitate the analysis of samples containing contaminants such as salts. Therefore, thin-layer preparations allow the avoidance of additional cleanup steps, e.g. with reverse-phase microcolumns purifications that are typically combined with dried-droplet HCCA preparations. The throughput can be significantly increased with such in situ purifications in larger 2D gel-based proteomics studies.21 The key observations using the Nd : YAG (Gaussian) on HCCA thin-layer preparations are the difficulties in selecting the appropriate laser power (‘criticality’) and the very quick sample ablation after only a few laser shots (‘sample consumption’). Figure 3 shows the peptide mass spectra obtained from identical HCCA thin-layer preparations with N2 (a) and with Nd : YAG (Gaussian) (b), both at comparable laser spot sizes of 50–60 µm (Figs 4 and 5(b)), as a sum of 20 shots. Better results for the Nd : YAG (Gaussian) can most likely be obtained from the thin-layer sample; however, in practice it takes a lot of trial-and-error experiments to find the optimal laser power. Since this is not feasible for every multisample plate, we consider the trace displayed in Fig. 3 as a typical result obtained with moderate tuning effort.

Figure 3.

Mass spectra of bombesin obtained with (a) N2, (b) Nd : YAG (Gaussian) on identical HCCA thin-layer sample prepared on a steel target, using the same instrument.

Figure 4.

Typical profile of the nitrogen laser (LTB, MNL 205 M/C), as observed in the experiments in this work. (a) View from top, (b) 3D view from side, (c) beam profile of eight consecutive laser shots, demonstrating the shot-to-shot variation in the profile of the nitrogen laser. Different colors represent different intensities.

Figure 5.

(a) Beam profile of a Nd : YAG (Gaussian) laser, fully focused. Measured FWHM is less than 5 µm, indicating the resolution of the CCD images; (b) defocused Nd : YAG (Gaussian) laser profile as used in the experiments. FWHM = 45 µm.

Since the performance differences shown in Fig. 3 were observed using the same sample in the same instrument, they have to be attributed to the different lasers. Physical parameters that may play a role can be wavelength, temporal beam shape or spatial beam shape. In an attempt to understand the phenomena, temporal beam shape was considered first, but was ruled out, as our experience using different nitrogen lasers with pulse durations between 800 ps and 3 ns had not shown any detectable dependence of laser threshold criticality and sample consumption on the pulse duration.

The wavelength of the laser was then suspected to be important for the observed effects, as the absorption curves of matrices as function of wavelength22 show that light at 355 nm has a higher optical penetration depth than at 337 nm. Therefore, higher fluence is required to reach the same energy deposited per unit volume, leading to desorption of more material and the behavior described above. In addition, as will be described below, the beam profile, even of the same laser, proved to be a lot more critical than the wavelength itself.

An examination of the spatial intensity distribution of the nitrogen laser beam at the sample target surface, the ‘beam profile’, showed that it is structured with an amplitude modulation by up to 50%. The maxima of the beam profile exhibited a minimum width of 10 µm. Most importantly this fine structure of the beam profile changed from shot to short! Further analysis was based on the assumption that the MALDI process critically depends on such needle-type local maxima and the typical Gaussian shape of the Nd : YAG focus is less favorable. To prove the hypothesis, we modulated the Nd : YAG beam profile from Gaussian to structured A or structured B. In these setups, using thin-layer preparations, more and better signals were obtained.

Analysis of the beam profile

A commonly used method to estimate the focal diameter is the size determination of the burn marks on a thin-layer matrix preparation using pulse energies well above the ablation threshold fluence, i.e. far above the fluences relevant in MALDI. However, in this case the Nd : YAG (Gaussian) created a clearly larger burn mark diameter than Nd : YAG (structured B), which was not consistent with their measured profile diameters. A second approach was to use a well-defined, analytically relevant pulse energy, e.g. 2 times the threshold fluence, for generating MALDI ions. Here, Nd : YAG (structured A) and N2 created no visible burn mark at all. The Nd : YAG (structured B) burn mark was barely visible and produced a circular mark with a diameter of 45 µm. Both burn mark approaches did not give any reliable data (data not shown); the results strongly depended on the thickness of the matrix layer and on the number of shots applied. Therefore, these ‘classical’ methods were not used in this work and all measurements rather relied on the CCD profiling of the laser beam in the focal plane of the MALDI target.

To enable the proper interpretation of the results from the MALDI measurements, single-shot beam profiles are described, which were used for the experiments. The Nd : YAG (Gaussian) produced very reproducible profiles from single shots, with a shot-to-shot variation within 3% of the overall pulse energy; thus single shots can be taken as being representative. The same is true for Nd : YAG (structured A or B); the character of the profile stays the same from shot to shot for a stationary modulator. In contrast, for N2 only the total pulse energy stays constant within 3%, but the detailed profile changes dramatically from shot to shot. Therefore each picture of an N2 beam profile can be seen only as an example that cannot be reproduced in subsequent laser shots.

N2 laser profile

Figure 4 shows different representations of the profile used in the experiments, a top view and a side view. Here the profile of LTB MNL 205 M/C is displayed, as this laser was also used for later experiments. We also examined the widely used LSI lasers, which show similar profiles (results not shown). Fitting a Gaussian curve to this particular shot yields a FWHM profile size of 52 µm × 75 µm. To visualize the shot-to-shot profile variation, Fig. 4(c) shows the profiles from eight different laser shots.

Nd : YAG (Gaussian) profile

Figure 5(a)) shows the beam profile of a focused YAG laser. The FWHM is less than 5 µm. This profile size was not used in the experiments, but it demonstrates the optical resolution of the setup. The low-intensity diffraction ring around the peak is due to the limited L2′ lens aperture. Figure 5(b)) shows a defocused Gaussian profile with a FWHM of 45 µm, which was used in the experiments described below.

Nd : YAG (structured A or B) profile

While many different profiles can be generated with the modulator setup, for simplicity only the two profiles shown in Fig. 6 were used in the experiments. One profile is large, termed ‘Nd : YAG (structured A)’, and the other one is smaller, termed ‘Nd : YAG (structured B)’. Fitting the two profiles with Gaussian shaped envelopes resulted in 60 µm and 45 µm FWHM diameters for Nd : YAG (structured A) and Nd : YAG (structured B), respectively. Besides the diameter differences, Nd : YAG (structured B) and Nd : YAG (structured A) profiles showed different substructures: Nd : YAG (structured B) consisted of nine well-separated Gaussian peaks of about 8 µm FWHM. Nd : YAG (structured A) is a more complex superposition of sub peaks. These sub peaks were slightly broader and denser compared to those in the Nd : YAG (structured B) profile. Figure 7 is a side-by-side comparison of all four profiles used in this work. It is clearly seen that the degree of modulation increases in the following order: Nd : YAG (Gaussian), N2, Nd : YAG (structured A) and Nd : YAG (structured B).

Figure 6.

Beam profiles of the two different structured beams used in this work. (a) Nd : YAG (structured A) with FWHM of a Gaussian fit of the envelope of 60 µm and (b) Nd : YAG (structured B) with FWHM of a Gaussian fit of 45 µm.

Figure 7.

Direct comparison of the four beam profiles used for the experiments.

MALDI performance

The main observations when using the Nd : YAG (Gaussian) for thin-layer HCCA preparations are (1) rapid sample ablation and (2) difficulties in setting the correct laser energy. For a description of these observations quantitatively, we measured the accumulated signal intensity for a given peptide (ACTH clip 18–39) as a function of the number of laser shots applied at a fixed sample position (Fig. 8). For each of the four different profiles, an optimized laser energy setting was selected that gave the maximum integrated signal intensity after 400 laser shots. After optimization of the laser energy, the acquisition was started on a fresh spot and the laser spot position remained unchanged throughout the whole measurement (400 shots). Packages of ten shots (Nd : YAG Gaussian) and 20 shots (N2, Nd : YAG structured A or B) were added. The actual content of the sum buffer was read out after each package and displayed as function of the total number of shots applied until then. The experiment was repeated five times, and the results were averaged.

Figure 8.

Integral peak intensity of ACTH clip (18–39) as a function of the number of shots taken from the same spot. The following laser powers were used: Nd : YAG (Gaussian) = 2.55 µJ, N2 = 0.78 µJ, Nd : YAG (structured B) = 0.23 µJ and Nd : YAG (structured A) = 0.14 µJ.

Early saturation was observed from the Nd : YAG (Gaussian) curve after 20 shots, and additional laser shots did not provide any more ion signals (Fig. 8). This was in accordance with the observation that the sample is used up after a few laser shots. The other three profiles do not show this early saturation and continue to contribute to higher intensity values of the peptide ions.

It is important to note the laser energy settings that were used in Fig. 8. They were set to maximize the integral signal from 400 shots: Nd : YAG (Gaussian) = 2.55 µJ, N2 = 0.78 µJ, Nd : YAG (structured B) = 0.23 µJ and Nd : YAG (structured A) = 0.14 µJ. Obviously, for Nd : YAG (Gaussian) much higher energy was used, leading to the smallest integral signal in Fig. 8. The requirement for higher Nd : YAG (Gaussian) laser energy was the reason for ablating away the sample so quickly. However, the product of the reached peak intensity integral and the used laser energy is constant within ±50% for all four profiles. Obviously, the validity of such a calculation is unquestionable, but a safe conclusion appears to be that the requirement for lower laser energy provides a larger MS signal. The different laser energies used in recording the data for Fig. 8 are illustrated in Fig. 9, where the profiles for Nd : YAG (Gaussian), Nd : YAG (structured B) and Nd : YAG (structured A) are stretched in such a way that the volume under the profiles scales with the applied laser energies.

Figure 9.

Relative laser energies of the beam profiles used in Fig. 8. The profiles are identical to the ones in Fig. 7 but stretched according to their absolute energy for illustration of the different energies used. (Color codes were not changed with stretching).

To identify the laser energy applied to thin-layer HCCA preparations for a ‘good’ spectrum, the resolution was plotted against the laser energy and normalized to the threshold laser energy (Fig. 10). To determine the threshold energy, in this work we used 1.2 times the threshold energy as the laser energy, which provided 10% of the maximum signal strength obtainable. Compared to the mere definition of threshold laser energy, observing 10% signal at 1.2 times threshold is more precise and easier to determine. For each point of the curve 200 shots were recorded on a new spot. The experiment was repeated three times and the results averaged.

Figure 10.

Mass resolution (2466 Da, HCCA thin-layer preparation) as a function of the relative laser energy (threshold laser energy = 1). The absolute threshold laser energies were: N2: 0.6 µJ; Nd : YAG (Gaussian): 1.7 µJ; Nd : YAG (structured A): 0.06 µJ; Nd : YAG (structured B): 0.15 µJ.

A steep decrease of the Nd : YAG (Gaussian) curve in Fig. 10 indicates that good resolution is obtained at threshold, but even a small increase in power already leads to a strong decrease in resolution. This corresponds well to the observation that finding the proper Nd : YAG (Gaussian) laser energy is difficult. The use of N2, Nd : YAG (structured A) and Nd : YAG (structured B) allows one to obtain mass spectrometric data close to the resolution maximum up to 1.7 times threshold energy, providing great comfort in setting the proper laser energy. Close to threshold, N2 provides slightly better resolution, while Nd : YAG (structured A or B) is least critical with regard to laser energies above 1.6.

In the same experiment we also recorded the integrated signal intensity as a function of the laser energy (sum of 200 shots, Fig. 11). Figures 10 and 11 quantitatively show the typical behavior of an HCCA thin-layer preparation: in terms of signal intensity with simultaneous resolution, Nd : YAG (Gaussian) is much worse. For the best reproducibility, the spectra for these figures were recorded in the central region of one single sample spot. In this area the maximum signal intensity reached was reproduced within 10% using a given laser energy. Resolution from a single measurement is reproduced in this area within up to 20% at threshold, up to 10% from 1.3 to 2 times threshold and around 5% at higher laser energy values. When different thin-layer preparations were compared (data not shown), the trend of Figs 8, 10 and 11 remains unchanged, with the differences in performance between Nd : YAG (Gaussian) and the other used arrangements increasing with decreasing preparation thickness.

Figure 11.

Signal intensity as function of relative laser energy (threshold laser energy = 1). The absolute threshold laser energies were: N2: 0.6 µJ; Nd : YAG (Gaussian): 1.7 µJ; Nd : YAG (structured A): 0.06 µJ; Nd : YAG (structured B): 0.15 µJ.

Figures 8, 10 and 11 show the major disadvantages of the Nd : YAG (Gaussian) on thin-layer HCCA preparations. The feeling of the operator that (1) the sample is ablated away very quickly and (2) it is so hard to find the correct laser power can be well rationalized. At the same time it can be seen from the figures that the use of a more suited laser focus profile really solves the problem. The performance of Nd : YAG (structured A or B) is quite comparable with that of N2 and is even better in some respects.

Other matrix preparations showed significant advantages using Nd : YAG (structured), e.g. the large crystals of DHB and 3-HPA preparations or sandwich preparations of SA. For example, we compared the results obtained with Nd : YAG (Gaussian) and Nd : YAG (structured A) on an oligonucleotide mixture prepared with the 3-HPA matrix. While Nd : YAG (structured A) gave intense, well-resolved peaks in the linear TOF MS, it was hard to achieve similar results with the Nd : YAG (Gaussian). When shooting with the Nd : YAG (Gaussian) onto large crystals at too low laser energy, no signal was observed, but when higher laser energy was used the crystals were blown off the target before detectable ions were produced. Finding a ‘sweet spot’ is much easier with Nd : YAG (structured) as compared to Nd : YAG (Gaussian). Even good spots for the Nd : YAG (Gaussian) do not give as good a signal as with Nd : YAG (structured). As 3-HPA sample preparation is very inhomogeneous, 100 shots on 17 user-selected spots were recorded, using an area of about 1/4 of the sample spot. None of the spectra was rejected on the basis of poor spectra quality, resulting in the accumulation of 1700 shots. One quarter of the preparation area was used for Nd : YAG (Gaussian) and another quarter of the same sample was used for Nd : YAG (structured A). Figure 12 shows spectra summed from 1700 shots each. Using Nd : YAG (Gaussian) it was hard to find a good signal at all, whereas using Nd : YAG (structured A), with a few exceptions, each set of 100 shots gave good signals. In this experiment, the laser energy was optimized beforehand on similar samples so that there was no need to optimize the laser energy again in the above experiment. The used laser energies were 17 µJ for Nd : YAG (Gaussian) and 1.4 µJ for Nd : YAG (structured A). It is important to note that some positions on the crystals did not work at all with the Nd : YAG (Gaussian) because these were ablated instead of providing the signal. In general, using the appropriate laser energy, the Nd : YAG (Gaussian) depletes much more material from the sample. This is shown in Fig. 13, which displays the sample before and after laser exposure. Obviously, with Nd : YAG (Gaussian), a lot more material was ablated as a function of the required relatively high laser energy. In addition to the quality aspect of the spectra, it is important to realize that such intense matrix ablation leads to strong contamination of the ion source. This is a serious problem especially in high-throughput applications.

Figure 12.

Linear MALDI spectra (1700 shots each) of the oligonucleotide mixture with 3-HPA matrix, about 50 fmole each. Trace (a) was recorded with Nd : YAG (Gaussian) and trace (b) with Nd : YAG (structured A).

Figure 13.

Oligonucleotide MALDI sample preparation with 3-HPA matrix as viewed through the observation optics of the instrument before and after spectrum acquisition (1700 shots): (a) Nd : YAG (Gaussian) before, (b) Nd : YAG (Gaussian) after, (c) Nd : YAG (structured A) before and (d) Nd : YAG (structured A) after.

When using SA as the matrix, the differences between Nd : YAG (Gaussian) and Nd : YAG (structured) were still visible, but smaller as with 3HPA. A dilution series of the BSA protein measured from 4 pmol down to 62 fmol showed that reasonable linear mode spectra were achievable with both lasers at all concentrations. Using a total of 250 shots accumulated on five different spots, Nd : YAG (structured B) was about 2 times better in S/N with a slightly better resolution (<10%) compared to Nd : YAG Gaussian (data not shown). Similar behaviors were also observed for other proteins. Typical laser energies used for BSA were 7 µJ for Nd : YAG (Gaussian) and 2 µJ for Nd : YAG (structured B). For in-source decay (ISD), MALDI-TOF measurements that provide sequence information directly from undigested proteins, (top-down sequence analysis),23, 24 the differences were larger. Figure 14 shows the result of a dilution series of RNAse from 10 pmol down to 1.25 pmol. At the highest concentration, the S/N for the ISD fragment ions acquired with Nd : YAG (structured B) is only twice as good as Nd : YAG (Gaussian). With the Nd : YAG (Gaussian), however, at lower protein concentrations the peak intensities faded more quickly, and basically no signal from ISD ions were detected in the spectrum from 1.25 pmol of RNAse. In all Nd : YAG (Gaussian) spectra the resolution is only about 50% of the resolution achieved with Nd : YAG (structured B). Laser energies used were 14 µJ for Nd : YAG (Gaussian) and 2 µJ for Nd : YAG (structured B).

Figure 14.

ISD (in-source decay) spectra of RNAse. For Nd : YAG (Gaussian) and Nd : YAG (structured B) four dilutions are shown: 10 pmol, 5 pmol, 2.5 pmol and 1.25 pmol. For each spectrum a total of 1000 shots were accumulated in packages of 100 shots from ten different laser spots.

Many questions remain for further investigation: Why is the Nd : YAG (Gaussian) operation noncritical on HCCA dried droplet, but critical with thin-layer HCCA preparations that create much smaller crystals? How can the size and distribution of the crystals cause such a performance difference? In addition, how do these results relate to the results from 3-HPA, where the crystals are of similar size or even larger than the laser focus itself? Why is the optimal fluence for Nd : YAG (Gaussian) so much higher than that for the same laser with a modulated beam profile? Does the efficiency of the MALDI process depend on the distribution of the fluence over multiple maxima rather than a Gaussian beam profile?


In this work, it could be demonstrated, for the first time, that using a structured beam profile instead of a Gaussian beam profile causes a frequency-tripled Nd : YAG laser to obtain analytically beneficial properties similar to a nitrogen laser. In some cases, the structured beam profile behaved even better than the nitrogen laser, possibly because the profiles used were less dense than those of a nitrogen laser. On some preparations that are typically advantageous for Nd : YAG (Gaussian), such as dried-droplet HCCA where small crystals are separated on the target support, the Nd : YAG (Gaussian) performance is comparable to N2 and Nd : YAG (structured). Therefore, certain applications can be run using Nd : YAG (Gaussian), but it is not an equal substitute for the nitrogen laser for general use. In contrast, the behavior of Nd : YAG (structured) is very close to that of N2 and therefore is a prime candidate for a general use laser. All work presented here is mainly relevant for axial MALDI-TOFs and may not be directly transferable to other mass spectrometers. MALDI is used around the world on thousands of instruments with many different applications. Therefore it will take a long time to prove that in any case, where the Nd : YAG (Gaussian) fails and the N2 works, Nd : YAG (structured) will work as well. But for those experiments performed to date, we have demonstrated that Nd : YAG (structured) worked much better than Nd : YAG (Gaussian) and sometimes even better than the N2. The applications that can dramatically benefit from the use of Nd : YAG (structured) include thin-layer preparations in LC-MALDI and gel-based proteomics, MALDI imaging, measurements from matrices such as DHB and 3-HPA (glycopeptides and carbohydrates) and protein profiling.

Finally, the strong influence of the profile structure adds a new parameter in the improvement of MALDI MS. At the same time, it raises new questions on the MALDI process itself, which may lead to a better understanding after more detailed and more systematic future research.


The authors would like to thank Anja Resemann, Martin Schürenberg and Detlev Suckau (Bruker Daltonik GmbH) and Franz Hillenkamp (University Münster) for fruitful discussions.