Single Particle Mass Spectrometry-Bioaerosol Analysis by MALDI MS †

The development of an aerosol mass spectrometer for the analysis of biological aerosols is described. The working principles of the aerosol mass spectrometer are particle sizing, particle selection and particle analysis by matrix-assisted laser desorption/ionization mass spectrometry (MALDI). The instrument has the capability of selecting only those aerosol particles that emit fluorescence when excited with 266 nm laser light, which differentiates the biological particles from virtually all other particles likely to be present in an aerosol. The implementation of a new ion source and delayed extraction has resulted in the capability of obtaining high-quality mass spectra of single bioaerosol particles. Isotopic resolution was obtained for a low-mass peptide. The sensitivity limit of the instrument was determined to be 1 zeptomole. The suitability of the aerosol mass spectrometer for the analysis of bacterial aerosol particles is demonstrated with an aerosol containing vegetative cells of the bacterium Escherichia coli when prepared off-line. The mass spectrum obtained has good resolution and covers a mass range up to 15 kDalton.


Introduction
The on-line chemical characterization of atmospheric (bio)aerosol particles has been investigated by researchers for many years 1,2) and has resulted in several (commercial) instruments.In 1988, Marijnissen et al. 3) proposed an instrument for on-line aerosol analysis in which size determination, laser-induced fragmentation and time-of-flight mass spectroscopy were combined.The novelty in the proposed apparatus of Marijnissen et al. 3) was the combination of particle sizing and laser fragmentation for time-of-flight mass spectrometry.Throughout the following years a substantial amount of research was carried out to optimize the performance of this aerosol mass spectrometer.For instance, Kievit 4,5) investigated the particle introduc-tion into the instrument and designed and characterized the aerosol beam generator.The aerosol beam generator is the interface which transports the aerosol particles from atmospheric pressure to the working pressure (10 -6 mbar) of the mass spectrometer.Weiss 6,7) implemented the aerodynamic sizing principle for the instrument.The aerodynamic sizing principle was adapted from Prather et al. 8) who were the first to implement such a sizing system in an aerosol mass spectrometer.Weiss 6,7) replaced the flash-lamp pumped Nd:YAG laser by an Excimer laser.Excimer lasers can be triggered much faster due to the shorter delay time between trigger and radiation output, and are therefore better suited for the application.In addition, with an Excimer laser, multiple wavelengths can be applied.Weiss 7) also implemented a new triggering circuit based on the design of Nordmeyer and Prather 9) , who were the first to implement such a triggering circuitry in an aerosol mass spectrometer.Software to acquire and process the data of the instrument was also developed by Weiss 7) .Stowers et al. 10) and Van Wuijckhuijse 11) implemented MALDI mass spectrometry, a technique that al-

Single Particle Mass Spectrometry−Bioaerosol Analysis by MALDI MS †
ⓒ 2008 Hosokawa Powder Technology Foundation lows the ionization of high-molecular-mass molecules and which is widely used for the analysis of complex compounds, including the analysis of bacteria [12][13][14] .
A fraction of the atmospheric aerosol particles are biological aerosol particles.A biological aerosol particle is defined as: an airborne solid particle (dead or alive) that is or was derived from living organisms, including microorganisms and fragments of all varieties of living things 15) .The microorganisms under consideration are viruses, bacteria, fungi, protozoa or algae.To reserve the instrument only for the analysis of the biological fraction of the total aerosol, Stowers et al. 16) implemented the collection of fluorescence emitted by biological particles, which was inspired by the work reported by, e.g.Pinnick et al. 17,18) , Hairston et al. 19) and Pan et al. 20) The amino acid tryptophan present in bacteria particles emits strong fluorescence when irradiated with 266 nm laser light.The emission of fluorescence is used as a selection criterion in the aerosol mass spectrometer.
All the above-mentioned developments have led to an efficient instrument which is able to generate (single particle) mass spectra of reasonable quality in terms of mass range and mass resolution of proteinaceous aerosol particles. 21)The performance indicates the suitability of the instrument for the analysis of bacterial aerosol particles.
Instruments capable of the on-line measurement and classification of airborne bacterial aerosol particles could be used for monitoring the atmospheric bioaerosol concentrations or in the defense against bioterrorism, but could also be used in hospitals or in hygienic production processes.
A few research groups are working on mass spectrometric instruments for the on-line analysis of bacterial aerosol particles.At the Lawrence Livermore National Laboratory, an on-line bioaerosol mass spectrometer (BAMS) is being developed.The applied mass spectrometric technique is LDI (Laser Desorption/Ionization) time-of-flight mass spectrometr y.This technique requires no reagents, and mass spectral signatures of individual spores are reported. 22,23)ergenson et al. 24) were able to discriminate individual spore particles of either Bacillus thuringiensis and Bacillus atrophaeus, based on the presence or absence of only one peak.
A more powerful approach for on-line bioaerosol analysis is aerosol MALDI mass spectrometr y, as employed by Murray and coworkers. 25,26)Jackson et al. 25) reported the on-line MALDI TOF MS analysis of proteinaceous material, whereby the particles were coated in-flight with the matrix nitrobenzyl alcohol.
Good results were obtained from an aerosol containing Escherichia coli with a semi on-line method.In this method, the bacteria particles were impacted on a MALDI target plate, which was precoated with matrix.This MALDI target plate was subsequently analyzed in a standard MALDI mass spectrometer. 26)esearchers at Oak Ridge National Laboratory use an ion trap mass spectrometer and have reported the on-line detection of proteins and peptides, applying the matrix by evaporation and condensation onto the aerosol particles. 27,28) o far, no on-line analysis of bacteria containing aerosol particles has been reported by the researchers from Oak Ridge National Laboratory.
On-line analysis of bacteria containing aerosols by aerosol MALDI has only been reported from the aerosol mass spectrometer in Delft. 10,11,21) towers et al. 10) reported the on-line aerosol MALDI MS analysis of Bacillus atrophaeus spores and found a reproducible peak at a mass of 1224 Dalton, which was attributed to a part of peptidoglycan.The work done by Van Wuijckhuijse 11,21) included the on-line aerosol MALDI analysis of spores of Bacillus atrophaeus and of vegetative cells of Escherichia coli.The obtained mass spectra covered a mass range up to 10 kDa.However, the S/R-ratio of the spectra was low and the spectra were hard to reproduce.This paper describes the working principle of the aerosol mass spectrometer and fur ther improvements of the aerosol mass spectrometer to improve the performance of the instrument in the analysis of bacteria-containing aerosol particles.The improvements include the implementation of a new ionsource and delayed extraction.The effect of the improvements on the performance of the aerosol mass spectrometer is discussed.The paper concludes with a description of analyses performed on off-lineprepared aerosol particles containing the bacterium Escherichia coli, to prove the suitability of the aerosol mass spectrometer for bacterial analysis.

Experimental
The aerosol mass spectrometer In Fig. 1, a schematic diagram of the aerosol mass spectrometer is given and is described as follows.
Aerosol is sucked into the instrument at a flow rate of 0.6 l/min through four differentially pumped chambers by a network of vacuum pumps.The pressure in the first section can be regulated and holds aerodynamic lenses which, together with the subsequent nozzle and two skimmers, act to generate a focused particle beam.The width of the beam was measured by Van Wuijckhuijse 11) at the location of the ionization laser focus (approximately 50 mm below the second skimmer) to be around 100 µm.The pressure in the ionization chamber is maintained at approximately 10 -6 mbar.The aerodynamic size of individual particles is obtained by measuring the transit time between two continuous laser beams, realized by splitting a randomly polarized, continuous-wave (cw) laser beam (532 nm) (MBD-266 laser Coherent Inc., Santa Clara, USA) into two beams with a beam displacement prism.The resulting 532-nm laser beams, focused to 75-µm spots, are 2.8 mm apart in a direction perpendicular to the particle trajectory.
In and at the exit of the nozzle-skimmer system, the particles are accelerated as a function of their aerodynamic diameter up to velocities as high as 250 m/s.A photomultiplier tube (PMT, Thorn EMI, Middlesex, UK; Type 9202B), located 45º relative to for ward scattering, records elastic scattering from particles intersecting the detection laser beams.The transit time is a measure for the aerodynamic size of the particle.The aerodynamic particle sizing is calibrated with monodisperse aerosol particles of polystyrene latex (Duke Scientific Corp, Palo Alto, CA, USA) in the size range of 0.24 to 5 µm.The density difference between natural aerosol particles and latex particles can be corrected by using eq.1: (eq. 1) In which da is the aerodynamic diameter, dp the particle diameter, ρp the particle density and ρ0 the standard particle density of 1 g/ml.The density of latex particles is 1.05 g/ml.
The alignment of the detection laser beams can be monitored using a digital camera (Canon Powershot A95) mounted at 90º relative to forward scattering.Preselection of the biological fraction out of the total aerosol is accomplished by collecting the fluorescence emitted from the biological par ticles.A part of the 532-nm cw-beam of the MBD-266 laser is deflected and goes into an external doubler, where a 266-nm laser beam is generated.The 266-nm cw laser beam intercepts the particle beam at the same spot as the top 532-nm laser beam.The emitted fluorescence is recorded on a gated UV-sensitive photomultiplier tube (Electron Tubes Ltd., Ruislip, UK; Model 9235QB) at 135º relative to forward scattering.A wave pass filter allowing light between 290 and 500 nm is placed in front of the photomultiplier tube.The collection time of the fluorescence by the PMT equals the residence time of the particles in the UV laser beam, which is approximately 1 µs 11,16,21) .
The time between the PMT output pulses is also used to trigger the Excimer ionization laser (ATL Lasertechnics, Friedrichstal, Germany; Model EC50), which is focused at 0.5 mm below the second detection beam.
The 308-nm Excimer laser, with a nearly flat topbeam profile and pulse width of 3 ns, establishes desorption and ionization of the particles.The UV laser spot area is approximately 300×500 µm and the laser fluence is adjusted for MALDI analysis, typically to values of approximately 1.5 J/cm 2 .As a result of the particle-size-dependent triggering and proper alignment, over 95% of all particles intersecting both cw beams are intercepted by the ionization laser beam to generate ions.This particle detection efficiency of the instrument was measured by comparing the particle concentration outside the instrument (using a particle counter) with the number of particles detected by the aerosol mass spectrometer during a certain time period at a fixed suction flow rate.The overall efficiency of the instrument is 1-5% for aerosol particles within the size range of 0.5 to 2 µm.
Positive ions produced in the ionization process are accelerated in a two-stage ion source toward a microchannel plate (MCP) detector that records the mass-dependent arrival time of the ions; this is called time-of-flight mass spectrometry.The two-stage ion source consists of a repeller plate, an extractor plate and a ground plate.The ions are formed in between the repeller and extractor plate and are accelerated due to the presence of an electric field.Further acceleration of the ions is caused due to the electric field between the extractor and ground plate.The length of the flight tube is 1.5 m.The voltages of the repeller, extractor, ion lens (to focus the ions), the deflector plates (to steer the ion beam) and detector are adjusted to obtain maximum response on the detector.
In the ion source, the ions formed are accelerated to the same final kinetic energy due to the presence of the electric field: where m is the mass of the ions (kg), v the velocity of the ions (m/s), e the elementary charge (1.6 10 -19 C) and V the acceleration voltage.In the drift region (the flight tube), no electric field is present and the ions formed cross the drift region with a velocity proportional to the square root of the mass of the ions: (eq. 3) The flight time (t) of the ions through the drift region of length L is then: (eq.4) Thus, singly charged ions of high molecular mass will have a lower velocity (eq. 3) and a longer flight time (eq.4) than singly charged low-mass ions.
The resolution of time-of-flight mass spectrometers is a result of three factors: the initial kinetic energy distribution, the spatial and the temporal distribution of the ions 29) .Each of these factors is discussed in the following.
Initial kinetic energy distribution is a result of the initial velocity of the ions.Ions with an initial velocity component along the time-of-flight axis will arrive earlier at the detector than same-mass ions with no initial velocity.Generally, the initial kinetic energy distribution is minimized by operating the source at high acceleration voltages.The relative difference between the smallest and the largest initial velocities decreases with respect to the final velocity due to the acceleration.The ion source in the aerosol mass spectrometer has been optimized to allow the use of high acceleration voltages, typically 35 kV.Spatial ion distributions are due to different positions of ion formation within the source, resulting in different travel distances in the source region and thus in ions with a kinetic energy distribution.Ions formed at the detector side in the source will enter the drift region sooner, but have a low velocity and will reach the detector later than ions formed higher up in the source region (further away from the detector).At a certain distance in the drift region the two above-mentioned ions catch up with each other.This distance is called the space-focus plane and is independent of the ion mass.Ideally, the detector should be located at the space-focus plane.A way to minimize the effect of spatial distribution is the use of a dual extraction source.The aerosol mass spectrometer is equipped with such a dual extraction source.
A dual extraction source consists of two extraction regions, of which the second extraction field is much higher than the first extraction field, where the ions are formed.
The temporal distribution includes the difference in time of ion formation and the device-dependent spread in ion detection due to the length of the flight tube for instance.Ions with the same mass, formed with a time difference δt, will enter the drift region with a time difference of δt and arrive at the detector with the same time difference δt.This time difference causes a lower resolution (R), which is deter-mined by: R = t 2∆t (eq.5) and for ions formed at time δt: (eq.5a) with ∆t at full width at half maximum (FWHM) of the peak.The FWHM (i.e.∆ t) is the width of a peak at the intensity, which is half of the maximum intensity of that peak.
Increasing the length of the flight tube, thus increasing the flight time, will increase the resolution.
Another way to minimize the temporal ion distribution is to change the way the ions are extracted.In the aerosol mass spectrometer delayed extraction is implemented.In delayed ion extraction, the voltages of the repeller and extractor plate are set to 35 kV.After a delay of typically 450 ns after ionization, the voltage potential on the first extractor plate is pulsed down to voltages of 30-25 kV to accelerate the ions.During the delay time there is a field-free region in the ion source and the formed ions distribute themselves according to their initial kinetic energy, as a result of the ion formation.
When the extraction field is applied, i.e. when the voltage of the first extractor plate is pulsed down, ions with high initial kinetic energy will arrive at the detector at the same time as the same-mass ions with low initial kinetic energy.When the initial kinetic energy of two ions is in opposite direction, the ion drifting away from the detector will receive more energy when the extraction field is applied than the ion drifting toward the detector.This results in a high resolution.
A 500-MHz digital oscilloscope (LeCroy, Chestnut Ridge, NY, USA; Model 9354CL) is used to sample the signals from the PMT and the detector and to send the spectrum output to a personal computer via a GPIB interface; the data are then further processed by a data system developed in-house.The instrument thus provides the aerodynamic size, fluorescence characteristics and a mass spectrum for all particles.Particles in the size range 0.2 to 10 micrometers can be analyzed with this system, however, the triggering circuitry can be adjusted to select a range of aerodynamic particle sizes or can be set to detect suitably sized particles with a fluorescence emission in the specified wavelength range.

Sample preparation
All chemicals, unless noted other wise, were obtained from Sigma-Aldrich Chemie BV, Zwijndrecht, The Netherlands, and were used without further purification.Protein particles were generated from solutions of analyte compounds pre-mixed with matrix using a Collison 6-jet nebulizer (BGI, Inc, Waltham, MA, USA).The flow rate through the nebulizer was set at 5 l/min, with approximately 0.6 l/min pulled into the instrument; the carrier gas was filtered air.The aerosol was dried in a diffusion drier packed with silica gel.Ten milligrams of the protein insulin or the neuropeptide substance P were dissolved in 5 ml of deionized water and the matrix materials 2,5-dihydroxy benzoic acid (DHB) or PMC, a proprietar ymade compound kindly provided by TNO Defence, Security and Safety, Rijswijk, The Netherlands, were dissolved in 3.3 ml of acetonitrile.Subsequently 1.7 ml of deionized water was added to the matrix solutions to dissolve the matrix compound.Just before aerosolization, the analyte and matrix solution were mixed so that the final solution had an analyte concentration of 1 mg/ml in deionized water/acetonitrile (2:1).The matrix/analyte molar ratio used was in the range of 400:1.Trifluoroacetic acid (TFA) was added to the insulin solution at 0.1 volume percent, to dissolve the insulin.Also, vegetative cells of the bacterium Escherichia coli K12 XL1 blue (cultured and harvested at TNO Defence, Security and Safety, Rijswijk, the Netherlands) were analyzed.Prior to nebulization, the bacterial samples containing 10 9 cfu/ml (determined by optical density at 260 nm), were washed three times by centrifuging the solution at 4000 rpm for 5 min-utes; each time the supernatant was removed and the bacteria pellets were then re-suspended in deionized water to approximately 10 9 cfu/ml.The solution was mixed with the same volume of a 20-mg/ml sinapinic acid matrix solution (SA) containing 0.1 % of TFA in a Petri dish.This mixture was vacuum-dried overnight and the resulting solids were ground.The produced powder was nebulized with a DeVilbiss powder blower (Model 175).

Improved resolution
An example of the increased mass resolution due to the implementation of delayed extraction is given in Fig. 2. The experiments were performed with aerosol particles consisting of insulin and the matrix PMC.First, continuous ion extraction was applied.In continuous extraction there is a continuous field between the repeller (35 kV) and extractor plate (typically 28 kV).This experiment was immediately followed by an experiment in which the ion extraction was delayed.Thus, the experimental conditions were the same, except for the different ion extraction methods.
The resolution is calculated according to eq. 5 for the main protonated insulin peak, at 5734 Dalton.As can be seen, the resolution of 900, obtained with delayed extraction (Fig. 2b), is higher than the resolution with continuous extraction, which is 350 (Fig. 2a).The insets in Fig. 2 show the appearance of the main protonated insulin peak at 5734 Dalton.The resolution in continuous extraction mode is too low to separate the sodium adduct.In single-par ticle spectra obtained in continuous extraction mode (data not shown), the sodium adduct can be distinguished.The inset in delayed extraction mode shows a peak attributed to the sodium adduct next to the main protonated peak.
The higher resolution in delayed extraction mode is due to the smaller difference in the flight times of the insulin ions.Note that the intensity of the peak is lower in the delayed extraction mode.The cause of this decrease in intensity is not clear.The implementation of delayed extraction has improved the performance of the mass spectrometer.A good resolution is required to make the instrument suitable for the analysis of high-mass molecules.

Isotopic resolution
Another way to examine the effect of the increased mass resolution due to the implementation of delayed extraction and the new high-voltage ion source is to investigate the presence of isotopes.For this purpose, an aerosol was produced from a solution con-taining the neuropeptide substance P (1347 Dalton) and the matrix PMC.In Fig. 3, a mass spectrum of a single particle (Fig. 3a) and a mass spectrum of a summation of 50 particles (Fig. 3b) is given.The inset in Fig. 3a shows the appearance of the main protonated peak, in which different peaks, with a mass difference of 1 Dalton, can easily be discriminated.The peaks are attributed to the isotopes, since the pattern corresponds to the isotope distribution observed in nature from this protein.The isotope distribution was obtained from http://prospector.ucsf.edu/prospector\newline/4.27.1/cgi-bin/msForm.cgi?form=msisotope.
The mass spectrum in Fig. 3b is a summation of the mass spectra of 50 single aerosol particles.Again, the inset shows the resolution of the main protonated peak.The different isotopes, as obtained for a single aerosol par ticle, are not clearly separated.Note that the single-particle resolution and the summed resolution are different.The resolution for a single particle is optimized by a high acceleration voltage and by delayed extraction, but due to the width of the particle beam, the place of ionization is not the same for each particle (spatial distribution), resulting in a  lower summed resolution.The spread in ionization locations could be decreased by reducing the particle beam width or by decreasing the diameter of the detection laser beams.However, this will result in a lower efficiency of the instrument.
The detection of isotopes in single-particle mass spectra indicates the applicability of the aerosol mass spectrometer for the analysis of bacteria-containing particles, since the resolution is so good that the individual components of more complex mixtures can be distinguished.

Sensitivity limit of the aerosol mass spectrometer
Identification of bacteria based on their MALDI spectra requires a ver y sensitive instrument.Madigan et al. 30) estimated the protein content of vegetative bacterial cells.The total number of protein molecules per cell is estimated to be 2.4 million and the number of different proteins is estimated to be around 1900.Therefore, the average amount of molecules per protein per cell will be in the order of 10 3 .Based on the above-mentioned numbers, a sensitivity requirement of 1 zeptomole (10 -21 mole) can be derived.
The sensitivity of the aerosol MALDI mass spectrometer is determined with an aerosol produced from a solution containing the protein insulin (5733 Dalton) and the matrix 2,5-dihydroxy benzoic acid (DHB).
The molar ratio of matrix to analyte in the solution was gradually increased from 500:1 to 50,000:1, by serial dilution.The solutions were aerosolized and the resulting aerosol particles are assumed to have the same matrix-to-analyte ratio as the original solution, assuming a homogeneously mixed solution.Up to a matrix-to-analyte ratio of 50,000:1, peaks caused by insulin ions were appearing in the mass spectra.
For this matrix-to-analyte ratio, the size range of the detected particles was changed from the whole detection range to the size range with an average aerodynamic diameter of 0.26 µm.In Fig. 4, an average spectrum of 30 0.26-µm insulin-containing particles at a matrix-to-analyte ratio of 50,000:1 is given.Russel et al. 31) determined the sensitivity limit of their BAMS system (bioaerosol mass spectrometer) and used eq.6 to calculate the number of analyte molecules per particle, assuming spherical, homogeneously mixed particles: (eq. 6) in which: and in which Nanalyte is the number of analyte molecules per particle, r the particle radius (cm), ρ the particle density (assumed to be the density of the matrix, which is 1.44 g/ml for DHB), MWmatrix the molecular weight of the matrix (which is 154 g/mole for DHB), NA Avogadro's number and da the aerodynamic diameter.Note that eq.7 is not correct.The aerodynamic particle size is defined as the diameter of a spherical particle with a density of 1 g/ml that has the same settling velocity as the particle.Formally therefore, the ratio of the densities should be used (see also eq. 1) 32) : in which ρ0 is the standard particle density of 1 g/ ml.Since the ratio of the densities has the same value as the density of the matrix only, the calculated values of Russell et al. 31) are correct.
Russell et al. 31) report a sensitivity limit of 14 zeptomole for their instrument.The obtained sensitivity of 14 zeptomole equals approximately 8400 molecules.If their BAMS system were to be applied for the analysis of bacteria particles, only a few ubiquitous types of proteins would be detected, covering a very small fraction of the protein content of a cell.
The sensitivity limit of the aerosol mass spectrometer is calculated by applying eq.6 and 8 for these experiments and gives a number of 600 molecules per particle, corresponding to 1 zeptomole.Although this value is lower than the sensitivity criterion defined above, it should be noted that this value is based on single-component aerosol particles.In a bacterium, many more types of molecules (biomarkers) are present, and it is highly likely that only a very few molecules of each biomarker will be detected by means of aerosol MALDI mass spectrometry.

Bacterial analysis by aerosol MALDI mass spectrometr y
The implementation of a new ion source and delayed extraction were inspired by the idea to optimize the aerosol mass spectrometer for the analysis of bacteria-containing aerosol particles.To investigate the suitability of the aerosol mass spectrometer for bacterial analysis, aerosol particles containing the bacterium Escherichia coli K12 XL1 blue and the matrix sinapinic acid were produced and analyzed.For comparison, the same culture of Escherichia was analyzed with a standard MALDI mass spectrometer (Biflex III, Bruker Daltonics, Bremen, Germany, located at TNO Defence, Security and Safety, Rijswijk, The Netherlands).The matrix material used was also sinapinic acid in water/acetonitrile 7:3 (v/v) 0.1% TFA.The sample was prepared according to the so-called dried droplet method.The obtained mass spectra are given in Fig. 5. Generally, the standard mass spectrum (Fig. 5a) is rather similar to the aerosol mass spectrum (Fig. 5b), with the remark that the resolution for the aerosol mass spectrum is better compared to the resolution obtained from the standard spectrum.However, more peaks at a massto-charge ratio higher than 17 kDalton are observed in the standard MALDI mass spectrum.Note that for standard MALDI analysis, resolutions are reported in the range of 10,000-25,000.
It can be concluded from this experiment with Escherichia coli that the mass spectrum from complex aerosol par ticles analyzed using the aerosol mass spectrometer covers a wide mass range and has good resolution.The highest mass peak appears at a mass-to-charge ratio of approximately 15 kDalton.This mass falls within the mass range (4-20 kDalton) proposed by Fenselau and Demirev 13) to contain the so-called biomarkers, which could be used for the classification of bacteria.However, since the highest mass detected by the aerosol mass spectrometer is lower than the proposed upper limit of 20 kDa, there is a chance that biomarkers are not detected by the aerosol mass spectrometer, making classification more difficult.Specially designed software to identify the peaks as well as matching the peaks with existing protein databases could be used for the classification of the samples analyzed by aerosol MALDI mass spectrometry.It can also be concluded from this experiment that the performance of the aerosol mass spectrometer for bioaerosol particles up to a mass of 15 kDalton is as good as the performance of the standard MALDI mass spectrometer which is used in this experiment, despite the differences in hardware of both systems.

Conclusions
The aerosol mass spectrometer as initially proposed by Marijnissen et al. 3) has been further developed and optimized to make the instrument suitable for the analysis of bacteria-containing aerosol particles.The implementation of a new ion source and delayed extraction has resulted in the capability of obtaining high-quality mass spectra of single bioaerosol particles.Isotopic resolution was obtained for a low-mass peptide on a single-particle level.The sensitivity limit of the instrument was determined to be 1 zeptomole for an aerosol of insulin.
The capability of the aerosol mass spectrometer for bacterial analysis has been demonstrated with an off-line-prepared aerosol containing the bacterium Escherichia coli.The obtained mass spectrum has good resolution and a mass range up to 15 kDalton is covered.However, more research is required to improve the sensitivity of the aerosol mass spectrometer to obtain good (on-line) single-par ticle mass spectra to allow discrimination on a single-particle level.

Fig. 1
Fig. 1 Schematic diagram of the aerosol mass spectrometer and a cross-section of the particle sizing and detection region.

Fig. 2
Fig. 2 MALDI mass spectra (summation of 1000 insulin particles premixed with the matrix PMC) obtained with a) continuous extraction and b) delayed extraction.The insets show the main protonated peak.

Fig. 3
Fig. 3 MALDI mass spectra of aerosol particles consisting of the protein substance P premixed with the matrix PMC; a) single-particle spectrum and b) summation of 50 aerosol particles.The insets show the protonated peak.

Fig. 4
Fig.4 Average MALDI mass spectrum of 0.26-µm aerosol particles of the protein insulin premixed with the matrix 2,5dihydroxy benzoic acid in a matrix-to-analyte ratio of 50,000:1.

Fig. 5
Fig.5 Mass spectra of Escherichia coli K12 XL1 blue with the matrix sinapinic acid; a) summed aerosol mass spectrum of 2000 aerosol particles (prepared off-line) and b) mass spectrum obtained with a standard MALDI mass spectrometer.