The potentials of MS imaging.

(1)

MSc Chemistry

Analytical Sciences

Literature Thesis

The potential of MS imaging.

An overview and recommendation

by

Joachim Isaak Zahradnik.

2170051

August 2019

Daily guidance:

Dr. F. Ariese

Second Examiner:

Prof. Dr. G. W. Somsen

(2)

2

Abstract.

This literature research centres around the basics, applications and developments of mass spectrometry imaging (MSI). This is a tool of analytical sciences that enables the measurement, detection and analysis of compounds in a spatial grid on a sample. In order to understand MSI, both the theoretical and practical properties will be explained.

Imaging is the principle of creating an image with the aid of measurements. If a measurement is done in a single dimension it is called 1 dimensional imaging (1D-imaging). With the inclusion of additional dimension this can be increased to 2-dimensional imaging (2D-imaging), 3-dimensional imaging (3D-imaging) and so on.

2D-imaging can be combined with mass spectrometry as long as the ion source is capable of performing a raster type of localized ionisation on a sample. Each single measurement is measured as any normal mass spectrometric spectrum and then logged as the input in a digital pixel. By combining these pixels and looking for specific values or intensities a two-dimensional image can be obtained from mass spectrometry data.

The mathematics behind the properties of MSI together with their actual correlations and the, often wrongly, various units will be explained along with their importance for MSI.

Mass spectrometry (MS) is a method of analysis that is one of the most basic tools for any analytical chemist. Although expensive, it yields data that are valuable to determine the properties, identity, quantity and cleanliness of a given sample. The three main components of a MS device: the ion source, the mass analyser and the detector. Of these only the ion source is of importance in regard to the capabilities of performing imaging measurements. The other parts influence the speed and quality of the single pixel measurements.

The different types of ion sources capable of performing an imaging type of mass spectrometry ionisation will be listed along with their functionality, applications and most importantly: a list of the progress of their imaging parameters as reported in papers over the past 30 years. The focus will be on the imaging part of the mass spectrometry instead of the mass spectra.

(3)

3

List of abbreviations.

1D-Imaging One Dimensional Imaging

1D-MSI One Dimensional Mass Spectrometry Imaging 2D-Imaging Two-Dimensional Imaging

2D-MSI Two-Dimensional Mass Spectrometry Imaging 3D-Imaging Three-Dimensional Imaging

3D-MSI Three-Dimensional Mass Spectrometry Imaging DESI Desorption Electrospray Ionization

ESI Electrospray Ionization

LA-ICP Laser Ablation Inductively Coupled Plasma LDI Laser Desorption/Ionization

LTPP Low Temperature Plasma Probe

MALDI Matrix Assisted Laser Desorption/Ionization MRI Nuclear Magnetic Resonance Imaging

MS Mass Spectrometry

MSI Mass Spectrometry Imaging m/z Mass to Charge ratio

Nano-DESI Nanospray Desorption Electrospray Ionization NMR Nuclear Magnetic Resonance

NIMS Nanostructure Initiator Mass Spectrometry SALDI Surface Assisted Laser Desorption/Ionization SIMS Secondary Ion Mass Spectrometry

SPCM Samples per centimetre SPD Single Pixel Duration SPI Samples per inch

(4)

4

1. Introduction.

Mass Spectrometry is one of the basic tools of any analytical chemist. It allows to obtain important data of the distribution of chemicals in any sample. By combining this with other analytical methods such as nuclear magnetic resonance (NMR), Raman, InfraRed spectroscopy and each of their imaging offspring. The obtained chemical data from the MSI can be combined to compensate for each other’s shortcomings.

MSI, the concept of analysing the entire surface of a large heterogeneous sample by repeated mass spectrometry analysis of parts of this sample, was introduced as a concept by Castaing and Slodzian in 19621. It took several decades before the concept was put into practice and the first MSI was performed. The invention of the MALDI ion source greatly aided this development as it created a simple way to immobilise a sample and locally ionize it without destroying larger molecules. Spengler and Kaufmann were the first to use MALDI to perform a MSI measurement in 19932.

The development of MSI took another boost when in 1997, Caprioli et al. applied the MALDI MSI to a biomedical sample. The ability to analyse peptides, proteins and other vulnerable molecules inside a slice of organ was a major breakthrough that would make it possible for analysts to identify the movement or location of specific markers, proteins and other important biomolecules at a cellular level.

The earliest papers on MSI came from biological, pathological and medical fields and focussed upon the analysis of larger molecules such as peptides and proteins.

Next the fundamentals of MSI will be discussed. From the principles on how one can perform imaging with any form of analytical device to the many properties that come with them. An explanation will be given on each of these properties and how they connect or are different from one another and how they are often incorrectly used. A useful table will be provided for those summarizing the main aspects.

Nowadays there are many more ion sources available capable of performing an MSI measurement. Both the mass spectrometer as well as the lateral (position of the sample grid) accuracy have improved greatly and these components are not showing signs of slowing down their rate of improvement.

The mass spectrometry part of MSI will start in Chapter 3, which will first explain the basics of mass spectrometry along with the mechanics and mathematics that are mandatory knowledge for any analytical chemist. Second, the method of obtaining imaging along with mass spectrometry will be explained. Thirdly, all ion sources capable of performing an imaging function and the developments they have gone through over the past years will be listed along with their applications. The focus will be on the development of the imaging capabilities of each ion source with some applications and mass spectrometry parameters as extra information. The advantages and disadvantages of each ion source will be listed as well along with a recommendation on which type to choose based on your sample.

(6)

6

2. Two Dimensional Imaging.

Two dimensional imaging is at its core the oldest form of analysis in existence. The mere act of looking at an object is to perform a two dimensional analysis of the reflection of light upon the surface of the sample. Over the past centuries the science has evolved and now we have obtained many forms of analysis that can analyse a wide range of properties on a two dimensional surface.

2.1. Basics of two dimensional imaging.

Based upon the principles of a digital camera one can swap the properties measured in a digital photograph (light wavelengths and intensity) with any property that can be obtained in a single measurement. Constructing all of these single measurements results in a full two dimensional imaging with a new measurement property.3

There is one major difference with the digital camera and all MSI methods. Where the digital camera obtains the picture in a single snap by measuring all the pixels at once, a mass spectrometry imaging instrument has to read the pixels one by one, unless one is somehow capable of obtaining and running as many mass spectrometers as there are pixels to be measured.

Figure 1 shows a good example on how the MSI measurements work.4 Firstly a sample is brought into position and a x-y grid is applied digitally for each measurement. The application of a x-y grid is called mapping. After running the MSI measurement the single spectra are applied to each section of the raster and a full image is digitally constructed based on the criteria of the scientist. Various images can be constructed from a single measurement by changing these criteria.

(7)

7

2.2. Applications.

2D-imaging is a form of analysis that goes back to when the first humans opened their eyes. In the countless aeons that humanity progressed since then their search for more ways to understand all of life’s many mysteries has intensified and their methods to perform imaging have become more and more sophisticated. To define the terminology of two dimensional imaging one could say that any form of observation was a measurement and each painting, drawing or description of such observations an indirect record of an imaging method. But, to keep things simple, only the real scientific imaging methods will be listed, starting with spectroscopy imaging. Which is the oldest form of imaging that relies on the enhancement of human sight with the aid of lenses. Even now it is so common that the device used for it, a microscope, is seen as the symbol of science in many modern media.

As science progressed, more ways of observing were discovered, developed and improved. In Table 1 a list is given of the imaging methods known. These are, for example, the use of long wavelength sound waves (sonar) to create a height map of the seafloors before any human reached them, the recording of deep infrared radiation to see stars never observed before or spotting the failure of isolation on the shell of a heat shield. As long as there is a way to perform many single measurements at once or to perform a sequence of them in a repeatable way on an unchanging sample, an imaging analysis can be made.

Table 1: Different types of imaging

Name of method Device(s) used Properties measured Applications Spectrometry Microscope, Camera. Distribution of light emission from surface of sample

Recording a visual observation to look at on a later date.

Sonar Echo locator Distance of

target to source

Dept measurements in water and surface of seabed.

MRI/NMR imaging

MRI scanner Distribution of

specific atoms

inside an object

Medical scans, detection of anomalies in atom distribution inside the subjects without damaging them

Thermal Imaging IR sensitive optics

Distribution of heat emission on the surface of an object

Security, thermal conduction and observation of heat generation or distribution. Radiography X-ray or Röntgen scanners Density of metals inside softer materials

Medical uses for detecting damages or deviations in bones. Security scans to detect hidden objects or persons inside of containers.

(8)

8

2.3. Mathematics of 2D-Imaging.

With so many different institutes, research companies and machines that occupy themselves with mass spectrometry imaging a lot of different settings are reported.

An important parameter that has to be discussed in MSI is the pixel count. This value is a true number that tells the number of measurements performed in a single MSI run by a whole numeral N. some articles list this as the “resolution” however the usage of this name for the pixel count is only for digital screens. Not for scientific measurements. Here the resolution is based on peak broadening and peak-to-peak separation.

𝑃𝑖𝑥𝑒𝑙𝑐𝑜𝑢𝑛𝑡 = 𝑁 (2.1)

The sample size is a simple m2 value that gives the size of the entire sample. There are a few studies in which the complete sample is analysed, while in other studies only a small segment of the entire sample is analysed.

𝑆𝑎𝑚𝑝𝑙𝑒𝐴𝑟𝑒𝑎 ≥ 𝐴𝑛𝑎𝑙𝑦𝑧𝑒𝑑𝐴𝑟𝑒𝑎 (2.2)

The Pixel Density, otherwise known as Samples per Inch (SPI) or Samples per centimetre (SPCM) is calculated by dividing the number of samples (pixels) by the total surface area that is analysed. But as we prefer to keep things metric we shall only be talking about the SPCM value in calculations. Take note that these are inverted squared values as the calculation is based on analysed area. Thus a squared value.

𝑆𝑃𝐶𝑀 = 𝑁

𝐴𝑛𝑎𝑙𝑦𝑧𝑒𝑑𝐴𝑟𝑒𝑎 (2.3)

SPCM is expressed in cm-2_{and can thus be inversed to obtain the unit we like known as pixel}

size. This value gives the surface that a single MSI measurement analyses.

𝑃𝑖𝑥𝑒𝑙 𝑠𝑖𝑧𝑒 = 1

𝑆𝑃𝐶𝑀= 𝐷𝑖𝑔𝑖𝑡𝑎𝑙 𝑃𝑖𝑥𝑒𝑙 𝐷𝑖𝑚𝑒𝑛𝑠𝑖𝑜𝑛𝑠 (2.4)

An important observation is that there is a difference between the true pixel size and the calculated pixel size, as most MSI measurements are not perfectly streamlined, there is either an overlap or a gap between two pixels during a MSI measurement. In some cases gaps are left on purpose to reduce the time needed for a single MSI run. Therefore, a distinction should be made between the calculated pixel size and the ‘true pixel measurement size’ that can only be obtained by studying the impact area created by a single measurement.4_{The true pixel}

measurement size depends fully on the impact size of the ionization beam/spray/cone of the ion source used.

(9)

9

The next important property for MSI measurements and settings: the time it takes. This too can be split up in several units and properties. First, the time it takes for a single measurement, single pixel duration. Second, the total measurement duration. The latter is the value that is important to companies that want to run an analysis and need to know the time it will take to perform.

Normally it would be simple to calculate the time by multiplying the single pixel duration (SPD) by the number of measurements taken. However, an MSI needs a short moment between measurements to change the position of analysis in the sample. These delays can add up. As Burger et al.5 have shown that in a 3 hour imaging run only a total of 83 minutes were actually spent measuring. This means that when trying to determine the SPD to replicate a method, oversampling can occur which either generates incorrect data or decreases the signal to noise ratio by introducing a larger background signal.

Unless the unit of Operational downtime is equal to zero the SPD cannot be calculated from the total measurement time nor can the total measurement time be predicted based on the single pixel duration. The Operational downtime is the sum of all time that is spent during the measurement when no sampling occurs. This could be either the time it takes for the ion source to heat up or the time between measurements at which the ion source is moved onto a new sampling spot or the sample is moved under a stationary ion source to a new sampling coordinate.

The total measurement time is an important value when working with samples that have a tendency to change. Such as life cells or decaying compounds. These require a low total measurement time.

𝑇𝑜𝑡𝑎𝑙 𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡 𝑡𝑖𝑚𝑒 = (𝑆𝑖𝑛𝑔𝑙𝑒 𝑃𝑖𝑥𝑒𝑙 𝑑𝑢𝑟𝑎𝑡𝑖𝑜𝑛 ∗ 𝑁) + 𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑎𝑙 𝑑𝑜𝑤𝑛𝑡𝑖𝑚𝑒 (2.6)

Finally there is a unit that normally is of little importance but can become relevant when one is working with valuable samples: the penetration depth of the ion source. This unit is usually left out of the settings as it is not so much a thing that can be controlled as a result of the ion sources properties.

It is an important indicator on how deep the ion source will ionize the sample. In cases where the sample is deemed of irreplaceable value, such as historical objects or pieces of art, it is preferred to have an ion source whose penetration depth is only a single molecule layer deep to minimize the damage that is done to these objects. There are claims of non-damaging ion sources but with the basic rules how mass spectrometers work these should be renamed ‘minimal impact ion sources’.

(10)

10

Table 2: Important properties for imaging using mass spectrometry

Name of property Units Explanation

Pixel Count X*Y or a numerical value Amount points measured in a single imaging run.

Sample size m2 Surface of the entire sample.

Analysed surface area m2 Surface area of the sample

that is analysed. Digital Pixel Dimensions

Or

Pixel Size

X*Y or r2π Area

Pixel size in the digital file

Practical Pixel Dimensions X*Y or r2π Area

Actual measurement size of the ion source.

Impact area of the ion source. Single pixel measurement

time

Time in s, m, u Time it takes for a single pixel to be measured or ionised. Total measurement duration Time in s, m, u Total time to perform the

entire MSI procedure

Even with Table 2 a clear indication should be made which of the properties are settings of tuneable devices and which are dependent on the materials used and thus cannot be altered. This will increase the reproducibility between labs if they can use the same materials as the publishing group did for their publication.

There is a big difference between the theoretical settings and the practical results that many reports fail to address. Many researchers will assume that several terms are connected and thus can be derived from each other without thinking about the imperfections of the machinery.

(11)

11

3. Mass Spectrometry Imaging.

3.1. Fundamentals of Mass Spectrometry.

A mass spectrometer consists of 3 important parts. The ion source, the mass analyser and the detector. The mass analyser separates gaseous molecules based on their mass to charge ratios. Its detector will measure the impacts of these molecules based on a variation in time or space (depending on the mass analyser used) and transform these values into mass to charge ratios (m/z). This means that for a mass analyser to work the molecules need to be brought into the gaseous phase and charged with either a positive or a negative charge. For this the ion source is used.

Ion sources come in many different forms and designs but their purpose is the same: to create gaseous or aerosol charged molecules out of a sample.

Two important distinctions can be made to catalogue the different ion sources. The first classification of types depends on whether or not the ion source requires sample pre-treatment. Think of putting the sample into a matrix or extracting the surface with solvents.

The second type of classification is made based on the type of energy that an ion source uses to evaporate and excite the target sample. These come mainly in 3 different types. The first type uses light or photons for excitation. The second type uses a solid particle bombardment such as a hydrogen beam for ionisation. The third type of ion source uses a liquid spray of charged solvent to obtain the sample molecules in a gaseous charged state.

For the importance of imaging purposes the mass detectors and mass analysers are of little influence. The key player to imaging is the ion source itself. Although the choice of mass analyser and detector will make a difference on the type and size of molecules you can analyse for the imaging MS. If you are using LA-ICP MS then it would be ill advised to use a TOF or orbitrap mass analyser as these have a low resolution for low mass ranges.

One of the properties that is of influence on imaging is the speed at which the mass separator and mass analyser can process a single spectrum (the time it takes to measure and to write away the data on the pc). In case of older devices this can become the bottleneck when attempting to measure at a high speed. Resulting in a long operational downtime between each ionization.

3.2. Mass Spectrometry 2D-imaging.

The combination of mass spectrometry with two dimensional imaging results in a method where one can take an MS spectrum of each ‘pixel’ in a two dimensional surface. This results in a high analytical solving power but a lower flexibility in imaging the sample as these are limited by the properties of the ion sources with mapping capabilities.

When Mass Spectrometry Imaging was introduced 20 years ago the main topic or sample used to be tissue for medical research. Soon enough the fauna researchers started getting interested

(12)

12

in this method as well and publications appeared on leaves and other plant material.67 Not much later many different types of samples were measured varying from serpentine urine in rock paintings to mummy wraps, fingerprints and banknotes.

3.3. Imaging out of MS.

To be able to perform a two dimensional imaging measurement with mass spectrometry one must be able to take located measurements or samples from a larger object of interest. This can be done with aid of either the ion source or by accurate sampling before introducing the sample to the ion source. As the second option allows any ion source to be used, the first option of ‘directly’ using an ion source in a two dimensional grit analysis is the better point of interest.8

There are several parameters of importance to these measurements. The two dimensional imaging segment can be seen as a grid like sampling matrix that holds various ‘settings’ that influence the MSI analysis. These settings and/or un-adjustable properties are of key importance to any form of imaging and should thus also be represented in MSI reports.

As shown in Figure 2 using the spatial flexibility of sampling as well as the ion sources of different mass analysis methods a two dimensional surface can be analysed in a single procedure.9 The procedure of obtaining spatial information regarding the composition of samples is an important part of structural analysis as well as a way to obtain information regarding the distribution of compounds of interest inside a biological sample. For large lateral resolution there is hardly any restriction on ion source for the mass analyser. The resolution for these simply depends on the method of sampling. Yet, once approaching the range of micrometres, the physical separation of a sample into smaller measurements is no longer feasible. More accurate measurements can only be obtained by improving the ion source.

(13)

13

4. Ion Sources suitable for MS.

4.1. Secondary Ion Mass Spectrometry.

Secondary Ion Mass Spectrometry (SIMS) is an ion source that functions by ‘sputtering’ a ray of charged ions (mostly metal ions) onto the sample. As shown in Figure 3 The high energy (primary) ion strikes the surface of the sample and transfers both energy and charge to create a secondary ion out of the inactive molecule that it collided with.6 This means that the ionization must take place under vacuum as impact with airborne molecules such as oxygen and carbon dioxide will make them lose energy or even transfer charge, resulting in a reduction of yield in ionization and an increase of impact area size.

Figure 3: simplified SIMS ion source diagram6

Figure 4: Mechanism of ionization propagation and fragmentation of ions. Wherein X+ _{is the}

(14)

14

Figure 4 shows how upon impact with the sample surface the ion’s energy and charge is transferred to the molecule it collided with. This generates an airborne and charged molecule that can be measured by the mass analyser. Figure 5 shows that often this impact and charge gives the organic molecule such a high energy that it becomes unstable and fractures during transition into the mass analyser, resulting in secondary fragment ions.10

The other possibility during collision is that the ion which collided with the molecule generates an organometallic compound with the target molecule. Due to instability from the added charge and excess electrical bonds these molecules will fragment to split off either the metal or a side group of the carbon atom to which the metal has bonded. 10

(15)

15

Table 3: SIMS Mass Spectrometry Imaging Reports

Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample. 1999 11 256 x 256 65536 0.6cm, 1.3cm 500 µm x 500 µm ~1 µm ~ 0.4 µm d 0.3 ms – 0.45 ms 20s – 30s Human Hair 2001 12 - - 200 µm x 200 µm - - 30 ns - Streptavin bound to substrate. 2002 13 7.9*10^4 15 µm radius 50 µm x 50 µm < 1µm radius 200 nm d 3.14*10^4 nm2 20 ns – 100 ns 1.6 ms – 8 ms Freeze-dried cells 2002 14 10^6 - 180 µm x 180 µm - 200 nm d 3.14*10^4 nm2 50 ns 0.05 s Freeze-dried cells. 2003 15 9*10^10 - 120 µm x 120 µm 0.16 nm2 - 20 ns 1800 s Polystyrene resins. 2003 16 <10^11 - 300 µm x 300 µm >0.9 nm2 _{100 nm d} 7854 nm2 30 ns < 3000 s Polypropylene film. 2004 17 128 x 128 16384 9 mm x 9 mm 500 µm x 500 µm 15.3 µm2 3 µm – 5µm d 7µm2 – 19µm2 200 shots at 5 kHz laser frequency. 0.04 s 655.36 s Freeze-dried brain sections 2005 18 256 x 256 65536 77 runs 1.73mm2 150 µm x 150 µm 0.34 µm2 _{500 nm d} 1.96*10^5 nm2 2.7 ms 180 s Freeze-dried nervous tissue 2005 19 256 x 256 65536 - 256 µm x 256 µm 1µm x 1µm 2 µm d ~3.14 µm2 7.3 ms, 9.6 ms and 91.5 ms 480 s, 630 s and 6000 s Freeze-dried brain sections 2007 20 256 x 256 65536 - 203 µm x 203 µm 790 nm x 790 nm 300 nm d 7.07*10^4 nm2 - - Freeze-dried aorta cuts. 2007 21 256 x 256 65536 - 256 µm x 256 µm ~ 1 µm2 _{100 nm d} 7854 nm2

Smaller than 30 seconds per scan. Unclear if this is per pixel or for the total image.

Lipid monolayer disposition 2007 22 256 x 256 65536 - 500 µm x 500 µm 1.95 µm x 1.95 µm 3µm d 7.07 µm2 - - High-pressure frozen adipose tissue

(16)

16 Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample. 2008 23 256 x 256 65536 5 mm x 6 mm 500 µm x 500 µm 1.95 µm x 1.95 µm 1-2 µm d ~4.9 µm2 150 µs 9830 s Cryostat tissue slices. 2008 24 256 x 256 65536 - 588.8 µm x 588.8 µm 2.3 µm x 2.3 µm

Sub-micron 25 ns 24 ms Snap-frozen spinal tissue. 2008 25 256 x 256 65536 2 mm – 2.5 mm 500 µm x 500 µm 1.95 µm x 1.95 µm

10 µm 0.16 s – 0.21 s 3 – 4 hours Plant tissue

2009 26 256 x 256 65536 2 inch d wafers 500 µm x 500 µm 2 µm x 2 µm 1-2 µm d ~4.9 µm2

< 2 ns pulse 0.13 ms Snap-Frozen liver tissue. 2009 27 256 x 256 65536 2 inch d wafers 500 µm x 500 µm 2 µm x 2 µm 1-2 µm d ~4.9 µm2 - - Snap-Frozen liver tissue 2010 28 256 x 256 65536 - 500 µm x 500 µm 1.95 µm x 1.95 µm ~ 400 nm d 30 ns 1.9 ms Snap-Frozen brain tissue 2010 29 250 x 250 62500 - 500 µm x 500 µm 2 µm x 2 µm 1-2 µm d - - Dogon painted stone. 2010 30 128 x 128 16384 - - 100 µm x 100 µm 0.61 µm2 3-5 µm d ~12.6 µm2 6.1 ms 100 s Peptide deposition on wafer 200 µm x 200 µm 2.4 µm2 _{24.4 ms} _{100 s} 2011 31 256 x 256 65536 1500 µm x150 µm x150 µm 500 µm x 500 µm 1.95 µm x 1.95 µm 1-2 µm d ~4.9 µm2 < 1 ns < 63.3 µs Painting sliver. 100 µm x 100 µm 390 nm x 390 nm 400 nm d 30 ns 1.9 ms 2011 32 128 x 128 16384 2 mm diameter 500 µm x 500 µm 4 µm x 4 µm 3 µm d - - Cryostat section of rat trachea. 2012 33 256 x 256 65536 1 µm thick 500 µm x 500 µm 2 µm x 2 µm 1-2 µm d ~4.9 µm2 < 1 ns < 0.26 ms Mummy skin 512 x 512 262144 200 µm x 200 µm 0.15 µm2 _{~ 400 nm d} 0.126 µm2 30 ns 7.8 ms

(17)

17 Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample. 2012 34 256 x 256 65536 3 mm thick 500 µm x 500 µm 1.95 µm x 1.95 µm 3-4 µm x 3-4 µm 5.18 ms – 13.7 ms 340 s – 900 s Mineral fragments. 2013 35 256 x 256 65536 - 500 µm x 500 µm 2 µm x 2 µm

2 µm d < 1 ns < 0.26 ms Vacuum dried eye tissue. 2013 36 128 x 128 16384 - 4 mm x 4 mm 976 µm2 - 512* 100 µs 51.2 ms 180 minutes Fingerprints on substrate. 2014 37 256 x 256 65536 - 500 µm x 500 µm 2 µm x 2 µm 1-2 µm d ~4.9 µm2 < 1 ns < 0.26 ms Snap-frozen lung tissue. 2014 38 256 x 256 65536 - 1200 µm x 1200 µm ~22 µm2 4 µm d ~12.6 µm2

4.1 ms 45 minutes Human skin.

2015 39 256 x 256 65536 1 mm x 1 mm 200 µm x 200 µm ~0.61 µm2 _- _{100 ns} _{6.55 ms} _{Breast tissue.} 2015 40 128 x 128 16384 - 100 µm x 100 µm ~0.61 µm2 - - - Bone cells. - 300 µm x 300 µm ~5.5 µm2 _- _- _- 2016 41 1024x1024 1048576 - 500 µm x 500 µm 488 nm x 488 nm - - - Amorphous and crystallised drugs. 256 x 256 65536 - 100 µm x 100 µm 390 nm x 390 nm

- 1 shot per pixel -

2016 42 2048x2048 4194304 - 500 µm x 500 µm 244 nm x 244 nm

70 nm x 70 nm 28.6 µs 120 s Sand and plastic particles. 2016 43 512 x 512 262144 - 100 µm x 100 µm 195 nm x 195 nm - 80 ns 20.9 ms Freeze-dried oocytes. 2017 44 256 x 256 65536 - 50 µm x 50 µm 200 nm x 200 nm ~ 300 nm d ~ 0.071 µm2

50 ms 1 hour Bacteria culture.

2017 45 512 x 512 262144 - 3 µm x 3 µm ~5.9 nm x ~5.9 nm - 3 ms 13 m Metallic nanoparticles.

(18)

18 Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample. 2018 46 128 x 128 16384 - 500 µm x 500 µm 3.9 µm x 3.9 µm Ca 3.5 µm 100 µs 1.64 s Resin cuts. 2018 47 768 x 128 98304 1 cm x 1 mm x 3 mm 3000 µm x 500 µm 3.9 µm - - - Wood cuts. 1024x1024 1048576 400 µm x 400 µm 390 nm 400 nm 100 ns 104.8 ms 2019 48 1024x1024 1048576 0.7 cm x 1 mm x 2.5 mm 400 µm x 400 µm 400 nm x 400 nm - 100 ns 104.8 ms Wood cuts.

(19)

19

Table 3 shows the development of the SIMS imaging methods published in the past 2 decades. The red italic values are derived by calculations from reported or calculated values. The red italic values in the column of ‘analysed surface area’ are often derived from measuring the picture supplied in the report along with the scale bar.

The publication of Gillen et al,11 in 1999 regards the analysis of a single, known, compound in human hair. The sample pre-treatment includes doping with organo-metallic solutions and mounting the hair upon conductive silver tape. The settings that they reported regarding the two dimensional imaging is complete and makes it relatively easy to duplicate.

The digital pixel dimensions of SIMS have not gone below 1 µm in size until Malmberg et al.20 publication regarding lipids in the aortic wall. Their publication was the first to get both digital and practical pixel sizes below the 1 µm dimensions. Even though other publications had indirectly proven to be able to measure such small dimensions they never indicated it in their papers. Unfortunately enough Malmberg et al. neglected to report their own measurement duration or single measurement time. Although this is a less important feature when one is trying to perform the best measurements, possible it is an important setting when trying to recreate the measurement performed, as well as when trying to balance the quality to quantity aspect of imaging measurements.

The latest article regarding SIMS MSI from 2019 by Fu et al.48 is not just about 2D-MSI but also 3D-MSI. This already shows the advancement in MS imaging over the past 20 years. In regards to the first article from 1999 all parameters have improved. The pixel size (both digital and practical) have decreased, single pixel measurement duration has decreased and the total pixel count has been increased as well.

When looking at the numbers there has been a steady rate of improvement on all factors. However, the improvements on practical pixel dimensions seem to have slowed down in recent years, indicating that little further improvement can be made as the boundaries of physics are met. As SIMS imaging is the oldest method for MS imaging the number of publications is impressive, up to about 1000 papers since the suggested method application in 1962 by Castaing and Slodzian.

Applications and practical usage:

SIMS as an ion source in 2D MSI is usable for a great variety of samples. most common of these are biological or organic samples such as slices of tissue, whole bacteria and leaves. Less common sample types are those that are not reproducible such as mummy tissue or paintings. For SIMS ionization little sample pre-treatment is needed. Although the sample must be able to fit into the sample holder for the two dimensional imaging. All in all SIMS is a flexible ionisation method that is easy to use. Its downside is the fairly large single pixel size and limited sample size.

(20)

20

Figure 6 shows an example of how a MSI measurement can give results for a single measurement.17 as the entire image is generated from various molecular masses measured in a single run. This shows the strength of recording an entire m/z spectrum as it allows one to generate a lot of different measurements without having to measure a sample for each compound of interest.

The data that is limited in the reports is commonly the theoretical pixel dimensions. Fortunately the other settings are not often omitted so there is a little chance in incorrect duplicate measurements. The errors that could occur are for example:

- Incorrect single pixel measurement time will lead to overexposure of the sample or underexposure in a single pixel

- Theoretical pixel dimension that is larger or smaller than the practical pixel dimensions results in parts of the sample being measured multiple times (with further decay on the overlapping spots) or spots that are left unmeasured and filled in by extrapolating between the actual pixel measurements.

- Omission of total measurement time can lead to some nasty surprises when trying to replicate the measurement. when a sample of 120 pixels with 1 second measurement time per pixel takes up 2 hours as the machine needs a minute to move between the pixel spots.

Figure 6: Negative TOF-SIMS images. (M - H)- denotes the quasimolecular ion corresponding to the entire molecule with a proton (H+) removed. The black horizontal lines (clearly visible in the phosphate image) are due to temporary loss of primary ion current during measurement.17

(21)

21

4.2. Laser Desorption Ionization.

With the observation that the impact of a laser of sufficient strength causes the target to fragment into fine, charged, particles, Cotter was the first who saw the opportunity to turn this new invention into an ion source49.

LDI is an ion source that uses a laser to locally ionize a sample. Thanks to the dual nature of photons to be both a particle and a wave this allows the transfer of both energy and charge in a focussed beam. This method does not require a sample pre-treatment if the target molecule is easy to obtain from the sample by laser desorption. The LDI only creates negative ions that have an added electron to their base weight, thus allowing direct mass measurement without having to take additives in mind. There are occasions in which the target molecule already had a multiple charged form that, upon fragmentation, creates both a positive and a negatively charged fragment.

The most important feature of LDI is the setting and properties of the laser used. If a shorter wavelength is used both the energy and the ionization rate increase. Furthermore, the chance of post-ionization fractions increases as well. For each sample a choice has to be made regarding the wavelength, duration and intensity of the laser pulse.

SALDI is based on the principle of MALDI. Although no matrix is used as the sample is simply adsorbed onto the surface of an active surface. The composition of this active surface can vary greatly and as a result there are many different types of SALDI with each their own surface material or surface properties that define the functionality and ionisation mechanisms.

Figure 7 shows how the surface material for SALDI is created.

Figure 7: Surface assisted LDI using nanosphere lithograph with Reactive Ion Etching (RIE).50

(22)

22

Firstly a pure silicon plate is oxidized to create a single molecule thick glass (SiO2) layer on top

of it. Secondly a SF6 solution is deposited on it and exposed to oxygen to generate small

nanoscale upstanding pillars on the surface of the silicon plate. The SF6 is then removed by

sonication along with the remainders of the glass layer so that only a pure silicon plate remains. As this is a random chemical process no two plates are identical. 50

According to Law & Larkin (2011) the different surfaces can be separated into three types of compositions and two types of functionality: Carbon based, semiconductors and metallic surfaces as compositions. With either electrical interaction or nanoscale as interaction types.

(23)

23

Table 4: LDI and SALDI Mass Spectrometry Imaging Reports

Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample. 2009 51 ? x ? 4633 - 0.46 mm2 _{10 µm x} 10 µm - 0.2 s 15 minutes Leaves. ? x ? 4125 0.41 mm2 14 minutes ? x ? 1107 0.11 mm2 _{4 minutes} ? x ? 9262 0.92 mm2 31 minutes 2010 52 ~60 x 50 ~3000 ~ 6 mm x ~ 5 mm ~30 mm2 _{100 µm x} 100 µm 75 µm d 2 s 200 minutes Leaves. 2010 53 18 x 14 252 900 µm x 700 µm 900 µm x 700 µm 50 µm x 50 µm

25 µm d 0.45 s 113 seconds Cortices cuts.

2013 54 1250x1500 1875000 ~12.5 mm x ~15 mm ~12.5 mm x ~15 mm 100 µm x 100 µm

60 µm d 0.25 s 130 hours Various cuts of different tissues. 2013 55 14 x 14 196 - 700 µm x 700 µm 50 µm x 50 µm

- 3 shots per step - Lipstick print on a sample tray. 2015 56 16 x 16 256 - 0.8 mm x 0.8 mm 50 µm x 50 µm

40 µm d 5 s 21 minutes Tumour tissue.

2015 57 ? x ? 14857 - ~2.14 mm2 _{12 µm x} 12 µm

10 µm d 0.5 s ~ 2 hours Epidermis tissue.

2015 58 50 x 50 2500 - 5 mm x 5 mm 100 µm x 100 µm - 0.48 s 20 minutes Fingerprints on a sample plate. 2017 59 ? x ? 24910 ~160 mm2 _{80 µm x} 80 µm

- 1.3 s 9.0 hours Cross sectioned

pencils and pencil stamps. ? x ? 15024 ~122 mm2 90 µm x 90 µm - 1.5 s 6.2 hours 2017 60 79 circular d of 100 1.5 mm d 1.5 mm d 1.76 mm2 150 µm x 150 µm

(24)

24 Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample. 2018 61 ~45 x 45 ~2025 - ~ 2.25 mm x 2.25 mm 50 µm x 50 µm

- 1.5 s ~51 minutes Metal coated target

plate. 2019 62 ~100 x 100 ~10000 Maize ~1 mm x 1 mm 10 µm x 10 µm

9 µm d

⅙

_s ~ 28 minutes Maize leaves.

- Corn - 100 µm x 100 µm 50 µm d ⅓ s - Corn leaves. 2019 63 10 x 10 100 - 500 µm x 500 µm 50 µm x 50 µm

(25)

25

Table 4 shows the development of both LDI and SALDI methods reported in literature. The red italic values are derived from values reported in the papers or calculated by hand based upon the values reported. The red italic values in the column of ‘analysed surface area’ are often derived from measuring the picture supplied in the rapport along with the scale bar.

The first paper regarding 2D-MSI with both SALDI and LDI dates from 2009 and was published by Hölscher et al.51 They performed multiple imaging measurements of different samples while keeping their settings identical apart from the matrix size and shape. In this paper they did a qualitative measurement to look for the presence and localisation of hypericins. This allows sacrificing mass spectrometry resolving power to improve the imaging power by decreasing its measurement duration and discarding all unwanted noise.

Most of these methods all feature some form of labelling of the sample, for instance the paper of Huang et al. from 201556_{. In this report they label mammal cells with a gold containing antibody}

before measuring the concentration of gold over the sample cells. Even though the number of pixels is relatively low (a 16 x 16 matrix for a total of 256 pixels) the total measurement time still adds up to at least 21 minutes per imaging measurement.

When looking at the progress of pure numerical values over the years it can be seen that on average the LDI methods have a relatively long pixel measurement time and thus are overall slow imaging methods. The biggest improvements are made in pixel size and the practical pixel dimensions that are in line with the decrease of laser spot sizes.

A remarkable feature is that several of the more recent papers describe methods where the sample is coated with an active metal layer such as silver or gold before being subjected to LDI. While improving the single pixel measurement time this method does add another step to the entire sequence of tasks that need to be done to go from a sample to a MS imaging method. This method is employed to improve the ionization rate thanks to the conductivity of the metals.

The sample pre-treatment for both LDI and surface assisted LDI is rather extensive and thus will induce a lot of error when trying to replicate it. Some samples are covered with a nanolayer of gold or silver while others require you to engrave a special silicon wafer to place a thinly cut slice of the sample on. This method is mostly used to measure larger molecules in complex organic matrices such as flower pedals and organ slices.

(26)

26

Figure 8. shows a LDI measurement of metal-coated maize seeds.62 The three different measurements each were on a resolving setting to analyse various components in the seeds. It also shows that there can be occasions where the sample pre-treatment has great influence on the measurements themselves. As the gold coating shows all of the analytes while silver results in missing half of the components that were looked for.

The main issue given with the data reporting of the papers is the lack of the practical pixel size (laser spot size) used in this experiment. In theory you could use any laser with a spot size smaller than the theoretical pixel size. Yet if you use the same wattage on the laser but the spot size is larger or smaller the intensity at the spot will be higher or lower than used in the original paper. Leading to sample decay/oversampling or insufficient energy to perform ionization of the larger molecules.

Another value that is lacking in the reports is the single pixel measurement time. This is a very important parameter as the lasers used are often of high power. Even with the pulse repetition rate it is key to know the number of shots fired at a single pixel during the ionisation. having this value wrong will once again lead to sample destruction or failure to ionise the target molecules.

Figure 8: Neutral lipids and some sugars in maize seeds with Au, Ag, and Pt metal matrices in positive ion mode. DAG diacylglycerol, TAG triacylglycerol. Scale bar = 1.5 mm62

(27)

27

4.3. Matrix Assisted Laser Desorption Ionization.

A common problem with LDI was that it destroyed the larger organic molecules. In 1986 the group of M. Karas et al. found a solution to this problem by immobilizing the sample in a matrix. The original goal was to reduce the stress upon the sample molecules but the workings were a bit different than intended. Fortunately this worked the way it was wanted and now MALDI is one of the most common ion sources.

Matrix Assisted Laser Desorption Ionization (MALDI) is a form of LDI where a sample has been immobilized by a matrix. There are several models for the exact mechanisms that happen during MALDI due to the complex interaction between the matrix and target molecules.

The functions are the same for LDI but the matrix increases the rate of ionization and allows normally non-volatile molecules to be ionized and charged by the laser source. MALDI usually uses a laser with a wavelength in the UV range for ionisation and is thus a rather hard ionisation source. The matrix contains molecules that have a strong resonance absorption at the wavelength of the laser used. This allows high energy lasers to be used as the target molecule will be shielded by the absorbing matrix that surrounds it. Figure 9 shows that the matrix molecules hit by the laser will absorb the photons and energy of the laser.64 This ionises them in such a rapid method that a small explosion occurs on the site of impact. This effect drags along the molecules that are trapped inside the matrix. Due to the effect of cluster forming the charged matrix molecules will cluster around the larger target molecules and grant them extra mass and, more importantly, a charge.

(28)

28

Table 5: MALDI Mass Spectrometry Imaging Reports

Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample. 1997 65 40 x 40 ~1600 1 mm diameter ~8x10^5 µm 25 µm x 25 µm ~25 µm d ~ 500 µm2

- - Piece of paper with

a copyright symbol. 2001 66 100 x 70 7000 12 µm 10 mm x 7 mm 70 mm2 100 µm x 100 µm - - - Tissue sample. 2002 67 400 x 400 160000 - 100 µm x 100 µm

250 nm - - 30 s Thin dye films on

aluminium foil. 2004 68 80 x 96 7680 - 4.0 mm x 4.8 mm

50 µm - 1 s 120 minutes Mouse liver and

brain slices. 2004 69 143 x 123 17589 28.5 mm x 22.5 mm 28.5 mm x 22.5 mm 0.2 mm x 0.2 mm 100 µm x 150 µm elliptical

2 s 9.7 hours Skin tissue.

2006 70 40 x 40 1600 - 12 mm x 12 mm

300 µm - 20 s 13.3 minutes Mouse livers.

2006 71 23 x 63 1450 - 0.25 mm x 9.45 mm 11.11 µm x 150 µm

- 9 Hz 161 s Rat brain tissue.

2007 72 ? x ? 598 - 23.92 mm2 0.2 mm x 0.2 mm 100 µm x 150 µm elliptical

2.5s 25 minutes Wheat stems.

2007 73 44 x 44 1936 22 mm x 22 mm x 0.15 mm - 0.5 mm x 0.5 mm 100 µm - 150 µm 400 ns – 800 ns 0.77 µs – 1.5 µs Various Tumor samples. 2008 74 ? x ? 12919 - - 100 µm x 100 µm - 500*100 Hz 5 s

~18 hours Brain tissue.

2008 75 146 x 133 19555 44 mm x 40 mm - 300 µm x 300 µm

- 2 s 10.8 hours Whole body tissue

sections. Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample.

(29)

29 2008 76 81 x 41 3321 - 8 mm x 4 mm 100 µm x 100 µm - 5 shots of a pulsed laser - Transfer Membranes. 2009* 77_example of bad math 75000 x 106 10 000 - 75 mm x 106 mm 1 µm X 1 mm 200 µm x 100 µm elliptical

1 µs 15 minutes Complete rat sections. 2009 78 63 x 63 3969 2.5 cm x 2.5 cm x 250 µm 2.5 cm x 2.5 cm 400 µm x 400 µm

- 500 shots - Peach cuts.

2009 79 ~38 x ~38 ~1444 ~7.5mm x ~7.5mm ~7.5 mm x ~7.5 mm 200 µm x 200 µm

- 5 s ~2 hours Various parts of a

sunflower. 2010 80 ? x ? 485 - ? x ? 4.85 mm2 100 µm x 100 µm 100 µm d 0.25 s 121.25 s Cell cultures on plates. 2010 81 333 x 250 83334 4 mm x 3 mm 4 mm x 3 mm - ~12 µm d 20 Hz laser 50 ms

69 minutes Plant leaves.

2010 82 112 x 80 8960 ~5.6 mm x ~4 mm x 15 µm ~5.6 mm x ~4 mm 50 µm x 50 µm ~150 µm x 75 µm elliptical

1 s 150 minutes Spleen sections.

200 x 200 40000 3 mm x 3 mm 15 µm x 15 µm ~150 µm x 75 µm elliptical 1 s 11 hours 2011 83 170 x 134 Or 114 x 200 ~22790 1.7 cm x 2 cm 1.7 cm x 2 cm 150 µm x 100 µm 150 µm x 100 µm elliptical 368 ms 140 minutes Condoms. 2011 84 250 x 200 50000 ~12.5cm x 10 cm ~12.5cm x 10 cm 50 µm x 50 µm - 200 shots at 200 Hz 1 second

~ 14 hours Mouse kidney.

2012 85 ~200 x 167 ~33334 ~12 mm x 10 mm ~12 mm x 10 mm 60 µm x 60 µm ~50 µm x ~150 µm elliptical

1 s ~9.25 hours Tumor tissue

sections. Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample.

(30)

30 2012 86 ~100 x 100 ~10000 ~2 mm x ~2 mm ~2 mm x ~2 mm 20 µm x 20 µm

10 µm d 1 s ~167 minutes Leaf section.

~286 x 58 ~16343 ~10 mm x ~2 mm ~10 mm x ~2 mm 35 µm x 35 µm

~272 minutes Root section.

2013 87 ~50 x 70 ~3500 ~12.5 mm x ~17.5 mm ~12.5 mm x ~17.5 mm 250 µm x 250 µm

150 µm d 1 s ~58 minutes Brain tissue.

~50 x 70 ~3500 250 µm x 250 µm 100 µm d 2⅔ s ~156 minutes 2013 88 ~70 x 200 ~14000 ~7 mm x ~20 mm ~7 mm x ~20 mm 100 µm x 100 µm

- 6.4 seconds 25 minutes Brain tissue.

2014 89 210 x 210 44100 - 2.52 mm x 2.52 mm 12 µm x 12 µm “reported” 5 µm

16.7 ms 12.25 minutes Fungi tissue.

2014 90 136 x 166 22576 0.9 cm x 1.2 cm 0.9 cm x 1.2 cm 100 µm x 100 µm 50 µm x 50 µm

⅓ second 125 minutes Lung tissue. 2015 91 150 x 300 Or 100 x 450 45000 - ~15 mm x ~45 mm 100 µm x 150 µm

30-40 µm 16.7 ms 125 minutes Leaf imprints.

2016 92 280 - ~2 cm 100 µm - 0.2 s 56 s Coronal sections. 2016 93 ? x ? 144682 - 130.2 mm2 _{30 µm x} 30 µm ~ 5 µm d < 40 ms 1.5 hours <96.5 minutes Rat brains. ? x ? 140772 - 126.7 mm2 30 µm x 30 µm ~ 5 µm d < 40 ms 1.5 hours <93.8 minutes ? x ? 24218 - 2.42 mm2 10 µm x 10 µm ~ 5 µm d ~37 ms ~15 minutes ? x ? 29236 - 2.92 mm2 _{10 µm x} 10 µm ~ 5 µm d ~31 ms ~15 minutes ? x ? 1015083 - 4.06 cm2 _{20 µm x} 20 µm ~ 5 µm d ~20 ms ~ 5.6 hours Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample.

(31)

31 2015 94 128 x 132 16896 - 9.6 mm x 9.9 mm 75 µm x 75 µm

- 1.9 s 9 hours Lung tissue.

214 x 160 34240 - 16 mm x 12 mm 75 µm x 75 µm - ~ 2 s 19 hours 2016 95 260 x 140 36400 - 7.8 mm x 4.2 mm 30 µm x 30 µm

5 µm d 1.3 s 13 hours Plant root.

360 x 140 50400 - 3.5 mm x 1.4 mm 10 µm x 10 µm 5 µm d 1.3 s 18.2 hours 2017 96 ~60 x 100 ~6000 - ~12 mm x ~20 mm 200 µm x 200 µm

- 1.25 s ~125 minutes Polymer film.

~15 x 30 ~450 - ~3 mm x ~6 mm ~9.35 minutes 2017 97 188 x 90 16920 - ~3.75 mm x ~1.8 mm 20 µm x 20 µm

30 µm d 0.2 s ~56 minutes Lung tissue.

2018 98 267 x 133 ~35556 ~ 2 cm x ~ 1 cm ~ 2 cm x ~ 1 cm 75 µm x 75 µm 50 µm d 0.2 s ~ 2 hours Fingerprints. 2018 99 ? x ? ~22000 - 107.8 mm2 70 µm x 70 µm

- ~1.3 s Ca. 8 hours Mouse liver tissue.

2018 100 10 x 11 110 - 1 mm x 1.1 mm 100 µm x 100 µm - - - Pollen Grains. 31 x 32 992 - 1.55 mm x 1.6 mm 50 µm x 50 µm - - -

(32)

32

Table 5 shows the development of the MALDI imaging methods reported in literature. The red italic values are derived from values reported in the papers or calculated by hand based upon the values reported. The red italic values in the column of ‘analysed surface area’ are often derived from measuring the picture supplied in the rapport along with the scale bar. While SIMS is older than the MALDI ion source the number of papers on MALDI imaging is astonishing. Many of these papers are of biomedical origin and do not report all of the important settings that an analytical chemist would be asking for when trying to replicate the imaging method.

The introduction of MALDI IMS in 1997 by Caprioli et al.65_{was a proof of concept where their}

measurement was simply the imaging of a copyright symbol that was printed on paper. They included the maths to remove the background of a matrix from the measurements to improve the signal to noise ratio. The developments of MALDI imaging as a whole clung closely to the improvement of lasers. Smaller spot sizes and higher energies from lower wavelengths allowed the pixel size to decrease along with the single pixel measurement time.

Around 2010 the first measurements with a pixel size below 100 µm started showing up. The lasers used from that point on can be divided into two categories. Either an elliptical shape that usually is 100 µm x 150 µm in size or a single round spot size that can go as low as 5 µm diameter. With this decrease in spot size also came an increase in pixel count on samples of the same size, resulting in larger total measurement durations.

Figure 10: Ion images of rat brain tissue with MALDI-TOF IMS. Observed substructures in the overlaid 30 μm ion image (A) of m/z 5024 (yellow), m/z 7070 (red), m/z 10 665 (blue), m/z 13790 (green) and m/z 15203 (pink) (B) is the trichrome staining of the same sample. High spatial resolution (10 μm) ion images were collected from the area outlined in black (C–E). The overall average spectrum for the 30 μm imaging experiment is shown in (F) highlighting each peak for the ions in (A).93

(33)

33

Figure 10 shows the effect of a MALDI imaging run in combination with a staining microscopy.93 This combination is greatly effective as it can allow one to have an overlay of where certain chemicals are located to support mass spectrometry imaging analysis. For the parts that do not contain any chemicals analysed will show up as black on mass spectrometry images.

Unfortunately enough in the past 2 years the amount of technical data reported has decreased in quantity as it appears that not many authors find it important to let the readers know the exact settings of their measurements.

When performing imaging mass spectrometry of a biological sample the go to method will be MALDI. Not because it is the best but because it is the most established method available. By immobilizing the sample in a matrix one is able to ionise molecules that otherwise would be hard to charge. The downside to this is the sample pre-treatment. You need to put your sample in a matrix and this introduces another level of error. As an upside it is easy to store your samples to measure again later on. Its complete measurement duration is not often directly given but can take a few hours on occasions.

The values that one has to derive from calculation (or which are hardly given) are those for practical pixel sizes, single pixel measurement duration and total measurement duration. Having to guess these can give trouble with planning of experiments as well as reproducing the measurement performed. An incorrect pixel size can lead to over/under-sampling of the sample. the lack of total measurement time can give rather nasty surprises when trying out a method as it reports a short single pixel time but omits its long operational downtime.

The use of the matrix protects the sample slightly against oversampling with the lasers and thus reduces the risk of molecule fragmentation by the excess energy slightly.

(34)

34

4.4. Desorption Electrospray Ionization.

DESI is an ionisation technique that combines the flexibility and softness of the electrospray ionization (ESI) ion source with the accuracy and localised sampling properties of the direct imaging methods. By turning a focussed ESI spray onto a sample it creates a fine mist of charged droplets that collide at high velocity with the surface of the sample.

By varying the settings shown in Figure 11 an optimized ionisation for DESI can be obtained.101

Alternatively, if the target is upon a dimensional moving platform (X and Y axis) a two-dimensional imaging can be performed.

The greatest strength of the DESI method is its flexibility in samples. As long as the target molecule can be dissolved in the charged solvent used in this method any object can be analysed in a 2D-imaging fashion. Before the development of wand or probe based ion sources an analyst was limited to samples which could fit inside the sample holder. With the invention of the DESI wand even large objects can have their surface analysed. However, the accuracy of this method for imaging is questionable as it has to be operated manually, introducing the biggest deviation in any form of analysis: the human error.

Nano-DESI is a variant of DESI where the solvent spray outlet has an internal diameter in the range of 50 µm. This allows for a much smaller area of impact and thus a far smaller pixel size on the imaging department of the MSI analysis. Although it is not as accurate as the laser based ion sources it still is an improvement over the accuracy of the conventional DESI ion source. A downside to this method is the fact that the mechanical parts of the setup have to be far more accurate to make optimal use of the decrease in pixel size.

(35)

35

Table 6: DESI and Nano-DESI Mass Spectrometry Imaging Reports

Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample. 2006 102 100 x 25 2500 - 10 mm x 10 mm 100 µm x 400 µm

- 1 second 42 minutes Thin layer plates.

2008 103 83 x 68 5644 - 45 mm x 47 mm ~540 µm x 0.7 mm - 5.4 s ~ 8.5 hours HPTLC cellulose sheets. 57 x 66 3726 - 31 mm x 49 mm ~540 µm x 0.75 mm - 5.5 s ~5.7 hours 2010 104 175 x 60 10500 17 mm x 8 mm 17 mm x 8 mm 200 µm x 200 µm

- 0.6 s ~105 minutes Adrenal gland.

35 x 20 700 7 mm x 4 mm 7 mm x 4 mm 5.1 s ~ 60 minutes 2010 105 100 x 100 10000 - 20 mm x 20 mm 200 µm x 200 µm 250 µm x 250 µm ~0.5 s 80 minutes Banknotes. 2011 106 40 x 40 1600 10 mm x 10 mm 10 mm x 10 mm 250 µm x 250 µm

- 1.5 s 40 minutes Human Seminoma

tissue. 2011 107 50 x 50 2500 - 10 mm x 10 mm 200 µm x 200 µm

- ~2 s 80 minutes Barley leaves.

2012 108 1100 x 15 16500 - 4.5 mm x 1.5 mm 5 µm x 100 µm 10 µm x 10 µm

- - Rat brain and

human kidney. 2012 109 40 x 20 800 - 10 mm x 5 mm 250 µm x 250 µm

- 0.16 s 128 seconds Rat brain tissue.

60 x 20 1200 - 15 mm x 5 mm 192 seconds 80 x 20 1600 - 20 mm x 5 mm 256 seconds 100 x 1400 140000 - 10 mm x 14 mm 100 µm x 100 µm - 0.31 s ~12 hours 2013 110 ~457 x 30 13700 - 8.2 mm x 4.5 mm 18 µm x 150 µm

- 1.2 s 5.2 hours Rat brains.

~220 x 38 8350 - 5.9 mm x 5.7 mm 27 µm x 150 µm - ~1.5 s 3.6 hours

(36)

36 Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample. 2013 111 152 x 273 ~41538 1.5 cm x 2.5 cm 1.5 cm x 2.5 cm 98.8 µm x 91.5 µm - 0.26 s 3 hours Fingerprints. 2014 112 ? x ? ~8926 - 557.9 mm2 _{250 µm x} 250 µm

- 1.21 s 3 hours Mice body slices.

? x ? ~56250 - 562.2 mm2 100 µm x 100 µm - 0.48 s ~7.5 hours 2014 113 125 x 30 3750 - 2.5 cm x 0.6 cm 200 µm x 200 µm

- 0.96 s ~ 60 minutes Zebra fish.

2015 114 320 x 15 4800 - 40 mm x ? 125 µm x ?

- 0.5 s 40 minutes Citrus Peels.

2015 115 ~40 x 25 ~1000 ~4 mm x ~8 mm ~4 mm x ~2.5 mm 100 µm x 100 µm - 540 ms ~9 minutes Leaf. ~35 x 30 ~1050 ~4 mm x ~9.5 mm ~3.5 mm x ~3 mm 100 µm x 100 µm - 540 ms ~9.5 minutes Leaf. 167 x 12 2004 10 mm x 0.7 mm 10 mm x 0.7 mm 60 µm x 60 µm - 540 ms 18 minutes Stem. ~60 x 180 ~10800 ~2 mm x ~11 mm ~3 mm x ~9 mm 50 µm x 50 µm - 543 ms ~98 minutes Leaf. 2016 116 130 x 100 13000 15 mm x 12 mm 15 mm x 12 mm ~115 µm x 120 µm

- 0.2 s ~ 43 minutes Vascular graft.

2016 117 ~625 x 50 ~31250 - ~15 mm x ~10 mm 24 µm x 200 µm - 0.6 ms reported, actual value is 0.6 s ~18.75 s reported ~5.2 hours true

Rat brain tissue.

2016 118 130 x 188 ~24000 75 mm x 52 mm 62 mm x 30 mm 480 µm x 3x160 µm

- 0.6 s 4 hours Mice whole-body

cut. 2017 119 450 x 170 76500 - 38 mm x 22 mm 80 µm x ~130 µm

(37)

37

Table 6 shows the development of both DESI and nano-DESI imaging methods. The red italic values are derived from values reported in the papers or calculated by hand based upon the values reported. The red italic values in the column of ‘analysed surface area’ are often derived from measuring the picture supplied in the rapport along with the scale bar.

Starting with the paper of Van Berkel et al. in 2006102 DESI MSI has been a method to quickly scan larger surface areas than the other MSI methods. Due to the large area that the nebulizer hits when performing a single pixel measurement this allows for an easy way to measure larger surface areas when compared to a laser based ion source. With the downside of having fairly long single pixel measurement times the DESI is a method that takes a while to complete an entire imaging measurement.

With the introduction of nano-DESI as an alternative in 2012 by Laskin et al.108 the pixel size greatly decreased in one dimension. As the thickness of the measured line became smaller it would still take some time before the displacement of the sample could be brought down to a matching level.

Even though the single pixel measurement times have decreased in the past 13 years DESI and nano-DESI have still relatively long complete measurement times.

The DESI ion source is an ion source employed to measure larger surfaces. Its relatively slow pixel imaging time along with the continuous measurement mechanism of the spray makes it capable of scanning lines with a low x-axial pixel size. The limitation to this is the fact that the practical pixel size is hardly ever recorded or mentioned in the papers. This is because the area that is targeted by the solvent spray will vary for every setup and thus the parameters to this setup are given instead.

(38)

38

DESI its large single pixel size makes it more suitable for measuring larger samples. it can be used to analyse the surface of any object that can fit the holder of the ion source. The incident angle, collection angle, tip-to-surface distance and MS-inlet-to-surface distance are also four important parameters to take into consideration when attempting to recreate these measurements. Sample pre-treatment is largely absent as the spray method is easy and quick but has a large practical pixel size with unknown exact size. Nanospray DESI has a smaller spot size and shorter single pixel measurement time. Otherwise, the methods are practically the same.

Figure 12 shows the process of sample preparation up to the point of a MSI result. With figure 13 being the resulting measurements of DESI-MSI on the fish along with an optical image of the fish in the sample tray.113 In general DESI requires little sample preparation other than immobilizing it in a mould.

The biggest issue with the DESI ion source is the flexible parameters of the ion source itself. As shown in figure 11 these will greatly influence the practical pixel size as well as efficiency of the ionisation.

Using too much solvent on a single pixel can result in a coating of the solvent slowly spreading out over the sample. But as long as one keeps this in mind the method is fairly fool-proof. Just keep in mind that nano capillaries used for nano-DESI are a magical world of problems on their own.

Figure 13: (a) Optical image of zebra fish in CMC mould. (b–f) Images of ion distribution produced from DESI analysis. (b) Image of deprotonated oleic acid m/z 281, (c) phosphatidylserine (PS 40:6) m/z 834, (d) phosphatidylinositol (PI 38:4) m/z 885, (e) bile salt 5α-cyprinol 27-sulfate m/z 531, and (f) sulfatide (ST 24:0) m/z 890.113

(39)

39

4.5. Nanostructure-initiator mass spectrometry.

NIMS is an ionization method that functions in a peculiar matter. NIMS substrates can be made of many sorts of materials, as long as they have nanostructural properties that can interact with the sample and the laser used for ionisation.

As the rules of nanoscale physics tend to behave in sometimes unpredictable manners, the predictions on efficiency of these surfaces is often a guess. However, their use lies most often in the detection and ionization of peptides and lipids as these large structures interact with the nanoscale structural features of the surface.

Figure 14: Example of NIMS sample pre-treatment on bacterial cultures.124

Figure 14 illustrates the sample pre-treatment for the NIMS ion source.124_{Here single cultures}

are first extracted using a gel which is used to transfer onto the NIMS chip. This procedure on its own will introduce an error to the dimensional properties of the sample due to the chemical interactions with the extraction gel.

(40)

40

Table 7: Nanostructure Initiator Mass Spectrometry Imaging Reports

Year of publication Total pixel count Sample size Analysed surface area Pixel Dimensions Practical Pixel Dimensions Single Pixel Measurement time Total Measurement duration Sample. 2009 120 - - - 100 µm x 100 µm

- 5 s - Adult mouse and

rat tissue. 2010 121 - 3.3 cm x 3.3 cm - 75 µm x 75 µm

- 20-50 shots - Mouse brains.

2010 122 - 3.3 cm x 3.3 cm - 75 µm x 75 µm

- 10-60 shots - Mouse brains.

2012 123 - - - 25-50 µm x 25-50 µm - 10 s - Mouse brains. 2013 124 - - - 50-100µmx 50-100µm - 0.09 s - Bacteria print on plate. 2013 125 ~110 x ~80 ~8800 - ~11 mm x ~8 mm 100 µm x 100 µm

75 µm d 0.1 s ~15 minutes Jonah Crab brain tissue.