Electrospray ionization mass spectrometry from discrete nanoliter-sized sample volumes


  • Patrik Ek,

    1. Division of Analytical Chemistry, Royal Institute of Technology, SE-100 44 Stockholm, Sweden
    Search for more papers by this author
  • Mårten Stjernström,

    1. Division of Analytical Chemistry, Royal Institute of Technology, SE-100 44 Stockholm, Sweden
    2. Division of Laserphysics, Royal Institute of Technology, SE-106 91 Stockholm, Sweden
    Search for more papers by this author
  • Åsa Emmer,

    1. Division of Analytical Chemistry, Royal Institute of Technology, SE-100 44 Stockholm, Sweden
    Search for more papers by this author
  • Johan Roeraade

    Corresponding author
    1. Division of Analytical Chemistry, Royal Institute of Technology, SE-100 44 Stockholm, Sweden
    • Division of Analytical Chemistry, Royal Institute of Technology, SE-100 44 Stockholm, Sweden.
    Search for more papers by this author


We describe a method for nanoelectrospray ionization mass spectrometry (nESI-MS) of very small sample volumes. Nanoliter-sized sample droplets were taken up by suction into a nanoelectrospray needle from a silicon microchip prior to ESI. To avoid a rapid evaporation of the small sample volumes, all manipulation steps were performed under a cover of fluorocarbon liquid. Sample volumes down to 1.5 nL were successfully analyzed, and an absolute limit of detection of 105 attomole of insulin (chain B, oxidized) was obtained. The open access to the sample droplets on the silicon chip provides the possibility to add reagents to the sample droplets and perform chemical reactions under an extended period of time. This was demonstrated in an example where we performed a tryptic digestion of cytochrome C in a nanoliter-sized sample volume for 2.5 h, followed by monitoring the outcome of the reaction with nESI-MS. The technology was also utilized for tandem mass spectrometry (MS/MS) sequencing analysis of a 2 nL solution of angiotensin I. Copyright © 2010 John Wiley & Sons, Ltd.

Electrospray ionization mass spectrometry (ESI-MS) has steadily gained importance during the past decades and is today an indispensable tool for identification and structural elucidation in bioanalytical chemistry.1 The soft ionization of the analyte molecules is a key feature for an extensive use in the analysis of non-volatile chargeable molecules such as proteins and peptides.2 An important improvement towards higher versatility was achieved when nanoelectrospray (nESI) emitters were developed.3 With these emitters lower flow rates and smaller initial droplets are generated than with the traditional electrospray devices, leading to an improved sensitivity.3, 4 Nanoelectrospray needles are now widely used both as column outlet for liquid chromatography (LC) and capillary electrophoresis (CE) as well as for offline analysis of individual samples.5

Current bioanalytical chemistry is to an increasing extent focused on clinical diagnostics and detection of early disease states of e.g. neurodegenerative disorders such as Alzheimer's disease, or cancer. However, the relevant biomarkers are often present in extremely low concentrations.6 This problem is usually approached by the use of immunoassays, combining specificity and high sensitivity, but assay development can be extremely tedious, and is not always successful. Detection usually involves fluorescence labeling, which can change the binding characteristics and natural activity of the sample, thereby affecting the analytical performance.7 Mass spectrometry does not have this drawback, since this is usually performed with non-labeled analytes. MS has the additional advantage of providing specific identification and structural information.8 Frequently however, the sensitivity of MS is not sufficient, and a concentration of the analytes into very small volumes is required.9 While sample volumes for matrix-assisted laser desorption/ionization (MALDI)-MS may be reduced to nanoliter size or below (particularly when an on-target sample concentration is performed),10, 11 the sample volumes required for nESI are much larger, typically on the order of microliters.12 The analyte concentration can therefore be critically low, when only a small amount of analyte is available. On the one hand, this stresses the demand for superior sensitivity of the MS instrumentation, which is a field of intensive development,1, 13, 14 but, also, it accentuates the great need for improved nESI-MS methods, where reduced sample volumes can be utilized.

In this paper, we describe a new technique to perform nESI-MS with small volumes of sample, taken up from the surface of a microchip. To minimize the risk for evaporation of the sample, we utilized a concept of covering the sample volume with a liquid lid of an immiscible fluid.15 The feasibility of the technology is demonstrated with the successful nESI-MS analysis of low-nanoliter volumes of protein/peptide solutions.



Acetonitrile (ACN) and methanol (both LC/MS, Chromasolv) were purchased from Riedel-de Haën (Seelze, Germany). Formic acid was from Fluka (Steinheim, Germany). Insulin (chain B, oxidized; bovine insulin; I-6383), angiotensin I (human acetate salt hydrate; A-9650), α-lactalbumin (bovine milk; L-5385), cytochrome C (equine heart; C-7752);, trypsin (T-8642) and ammonium acetate (A-7330) were purchased from Sigma-Aldrich (Steinheim, Germany). Liquid fluorocarbon (FC-77 Fluorinert™, boiling point 97°C, ρ 1780 kg/m3, µ 1.3 mPas) was purchased from 3M Company (St. Paul, MN, USA). The water used was purified with a Synergy 185 system from Millipore (Bedford, USA).

Generation of small sample volumes

A schematic of the setup utilized for generation of small sample volumes is shown in Fig. 1(a). The entire equipment was mounted onto an optical table (TMC, Peabody, MA, USA). A silicon chip (ca. 1 × 1 cm) was placed on the bottom of a glass cuvette (3 × 10 × 1.5 cm; w × l × h). The cuvette was filled with approximately 20 mL of the fluorocarbon liquid. One end of a ca. 25 cm long fused-silica capillary (50 µm i.d., 375 µm o.d.; Polymicro Technologies, Phoenix, AZ, USA) was etched to a pointed shape in hydrofluoric acid.16 The blunt back end of the capillary was passed through a rubber membrane into a Pyrex glass vessel (1 mL) which contained a plastic Eppendorf tube (Protein LoBind Tubes, Hamburg, Germany) with the sample solution. The blunt capillary end was immersed into the sample solution. The front end of the capillary was mounted on a XYZ translation stage (Eksma, Vilnius, Lithuania). By means of the translation stages, the pointed end of the capillary was aligned approximately 50 µm above the silicon chip. A 1 bar pressure pulse was applied to the glass vessel via an electrically controlled valve. This resulted in the formation of a droplet of sample at the pointed end of the capillary. After the droplet had made contact with the silicon surface, the capillary was pulled up with the aid of the translation stage to release the droplet. The volume of the deposited sample volume could be controlled by the duration and magnitude of the pressure pulse. To avoid cross-contamination, a new deposition capillary was employed for each new sample solution.

Figure 1.

(a) Schematic of the setup employed for deposition of small volumes of sample onto the microchip. (b) Schematic of the aspiration of the sample into the nESI needle. (c) A series of photographs of a sample volume of 1.8 nL that is aspirated into the needle tip.

Sample aspiration into the nESI needle

Throughout the experiments, two types of nESI needles were used; Borosilicate glass tips (ES380, Proxeon, Odense, Denmark) and quartz tips (QuartzTips™, QT10-70-2-CE-20, New Objective Inc., Woburn, MA, USA). A new needle was used in every experiment, unless stated otherwise. A schematic of the setup is shown in Fig. 1(b). The rear end of the nESI needle was connected via a drilled hole in a rubber septum to a gas-tight glass vessel (ca. 2 mL) with dual-sided threads. The air vessel was mounted on a XYZ linear translation stage. The tip of the nESI needle was aligned above a deposited sample volume under the fluorocarbon liquid lid using the translation stages. The glass vessel was connected to an electrically controlled valve via a Teflon tube. The valve was connected to water vacuum suction through an additional Teflon tube. A negative pressure (ca. 0.8 bar) was applied to the air vessel by switching the valve. First, a small plug of liquid fluorocarbon was aspired into the needle. Subsequently, the needle was immersed into the sample solution volume and aligned as close to the silicon chip surface as possible by means of the translation stage to aspire the sample solution. Finally, an additional small plug of fluorocarbon liquid was aspired into the needle before the negative pressure was released by switching the valve to its original position. A series of photographs in Fig. 1(c) shows the aspiration of a sample volume into the nESI needle. After removal from the setup, the needle, filled with sample, was rapidly mounted in the nESI interface of the mass spectrometer.


A CM-10 microscope (Nikon, Tokyo, Japan) equipped with a CCD camera (model CX-ES30CE, Sony, Tokyo, Japan) and an ultra-long working distance objective was employed to visualize the deposition of sample solution as well as the subsequent aspiration of the sample solution into the nESI needle. An additional CCD camera (model C2400-75i, Hamamatsu Photonics, Japan) with a magnification objective was positioned at 90° with respect to the microscope. This arrangement allowed an exact positioning of the deposition capillary and the nESI needle. The volumes of the sample droplets were calculated as segments of spheres by measuring the dimensions in photographs of the samples with the computer software ImageJ.17 The estimated volumetric error was less than ±6%.

Electrospray mass spectrometry setup

For the MS analysis, an ion trap mass spectrometer (ESQUIRE, Bruker Daltonics, Bremen, Germany) was used. The mass spectrometer was operated in positive ESI mode, with normal scan resolution (13000 m/z/s) using the software EsquireControl (Bruker Daltonics). The applied voltage for the nESI was between 450–600 V. The temperature at the MS inlet was 30°C. No nebulizing pressure was applied. Evaluation of data was carried out with the software DataAnalysis (Bruker Daltonics). A single mass spectrum was based on an average of five individual mass range scans with accumulation times adjusted by setting the ion charge control to 50 000.


Some of the fundamental ideas of miniaturizing analytical systems are to improve sensitivity and speed of analysis, as well as to reduce sample consumption and cost.18 In bioanalytical chemistry, the issue of increased sensitivity and decreased limits of detection has become of prime importance, particularly in clinical applications dealing with detection of the early stages of diseases.14 As pointed out in the introduction, mass spectrometry has a number of attractive features compared to immunological methods. Frequently, however, the concentration of relevant analytes may be as low as in the femtomolar range and the available amount of biological sample is often limited, which means that the amount of analyte and the concentration after sample work-up may be too low to allow detection by MS. In such cases, a further concentration down to extremely small volumes may be necessary. However, the degree of volume reduction must be in parity with the minimum volume, which can be handled by the system. For the conventional nESI technology, these volumes are typically on the order of microliters. Recently, digital microfluidic systems have been utilized on-line with nESI-MS. These efforts have demonstrated the feasibility to perform electrospray with very small sample volumes.19, 20 Furthermore, aspiration of minute volumes from a single cell through the tip of a nESI needle with additional dilution into a microliter volume of electrospray solvent prior to MS analysis has been presented.21 However, for ultra-trace components, obtained from comparatively large volumes of biological material, off-line handling is necessary and subsequent nESI-MS analysis of very small volumes of concentrate remains an unsolved issue. Our previous work, dealing with a new sample handling setup for microarray assays,22 inspired us to develop a new method for handling and nESI-MS analysis of very small sample volumes. With this setup, picoliter-sized sample volumes can be generated, which are deposited as small droplets on a flat silicon microchip. A fundamental problem when handling such small volumes of sample is evaporation.23 From model experiments, we observed that a sample volume of 3 nL (50:50 (v/v) acetoniltrile/water) evaporated in about 10 s. To solve this problem, we used a system where the sample was never exposed to the open air. All sample manipulation steps, including the deposition of the sample droplets on the microchip and the subsequent transfer of the sample into the nESI needle, were performed under a cover of Fluorinert™ FC-77, which is a volatile and inert fluorocarbon liquid of low viscosity.15 Under these conditions, the loss of liquid from a sample volume of 3 nL at room temperature was only approximately 5 pL/min. By using precision micromanipulators, more than 95% of the sample volume could be transferred into the tip of the nESI needle, even for sample volumes as small as 1 nL. In order to avoid evaporation of sample from the needle during transfer to the electrospray setup, the sample plug was sandwiched inside the needle by two small plugs of liquid fluorocarbon.

In an initial experiment, a series of nine samples containing angiotensin I was analyzed in consecutive order, using a borosilicate glass nESI needle. The sample volumes ranged from less than 2 nL to 50 nL. We utilized one and the same ESI needle for all MS runs to facilitate a quantitative comparison of the results. The entire sample volumes were utilized, and the signals of the extracted ions for the masses corresponding to singly, doubly and triply charged angiotensin I were integrated. The electrospray duration for the analyses ranged from 0.1–1.8 min. Figure 2 shows the integrated total intensities, plotted against the sample volumes. The expected linear dependence of the signal as a function of sample size was confirmed, which shows that the setup is robust for multiple, quantitative analyses, using one and the same emitter.

Figure 2.

Graph showing the total integrated signals of the extracted ions for the masses corresponding to singly, doubly and triply charged angiotensin I of nine samples with different volumes. (The data points from the experiments with the 2.6 nL sample volume and the 2.9 nL sample volume partly overlap in the graph).

The smallest sample volume which could be analyzed was 1.5 nL. A spectrum of such a sample is shown in Fig. 3. The main limitation for analyzing smaller volumes is the short duration of the electrospray. With needles having a smaller orifice, it should be possible to generate lower flow rates, and in this way, it should be possible to obtain an increased sensitivity.3, 4, 24, 25 The borosilicate glass needles which were used had a specified orifice dimension of 1–2 µm, and more exact figures for upper and lower tolerances could not be provided by the manufacturer. It was even stated that the tip of the needles might be completely sealed, and it was therefore recommended by the manufacturer to open the needles by gently breaking the tip before usage. It was therefore not possible to obtain a reproducible, low flow rate, when using different needles.

Figure 3.

ESI mass spectrum generated from a 1.5 nL sample volume containing 10 µM angiotensin I (MW 1296.5 Da) in 49.5:50:0.5 (v/v) water/methanol/formic acid.

In subsequent experiments, we compared the performance between nESI needles fabricated in borosilicate glass or quartz. We noted that, on average, the borosilicate glass needles generated lower flow rates than the quartz needles, based on the observed duration of the electrospray generated from known sample volumes. The orifice diameter of the quartz needles were, according to the manufacturer, 2 ± 1 µm, i.e. slightly larger than the borosilicate glass needles, which was also confirmed by SEM measurements. Additionally, as shown in Fig. 4, the borosilicate glass needle tips had the shape of a long elongated trunk, while the quartz tips narrowed into the tip region with a shorter tapered elongation. The longer tapering is anticipated to increase the flow resistance.26 The lower electrospray flow rate generated with the borosilicate glass needles should result in higher sensitivity in comparison with the performance of quartz needles. Nevertheless, a significantly better signal-to-noise (S/N) ratio was observed when using the quartz needles. This is shown in Fig. 5, where ESI mass spectra from 100 nM α-lactalbumin obtained with a quartz and a borosilicate needle are compared. The spectra represent the collected signals from the entire sample volumes (40 nL for the quartz needle and 80 nL for the borosilicate needle). The superior results obtained with the quartz needle are likely due to a lower analyte adsorption.12 The lowest detectable concentration was obtained for a sample solution of 10 nM insulin (chain B, oxidized) using a quartz needle. A sample volume of ca. 80 nL had to be utilized in order to obtain an electrospray for a sufficiently long period of time. Thus, the integrated signal corresponds to an absolute amount of 800 amol of the insulin. Since significantly lower flow rates could be obtained with the borosilicate needles, it was possible to run the nESI with much smaller sample volumes: 2.1 nL of a 50 nM solution (corresponding to an absolute amount of 105 amol) yielded acceptable data. This amount of analyte would correspond to a concentration of 210 pM in a conventional nESI analysis (0.5 µL volume). In a comparative test with conventional sample loading, where we used such a concentration of analyte, we could not detect any signal, even after spraying the entire 0.5 µL of sample, and integration of the signal obtained from the 25 min long electrospray process. The results obtained from the comparative experiments are shown in Fig. 6. Thus, it can be concluded that concentrating and confining the sample solution into a small volume results in a significant increase in sensitivity and detectability. It should be possible to obtain a considerable additional improvement of the limit of detection by using a more modern MS instrument, including better ion transfer techniques such as ion funnels, etc. The experiments also showed the importance of the dimensions and material of the needle. It seems reasonable to assume that still lower absolute detection limits could be obtained by using quartz needles with very small orifice dimensions. We are currently investigating this.

The open access to the deposited samples on the flat silicon surface provides the interesting possibility for the droplets to act as isolated compartments in which chemical reactions can be performed. This is shown in an experiment where we carried out an enzymatic digestion of a protein. Two separate sample volumes (15 nL each) of cytochrome C in 10 mM ammonium acetate (pH 8) were deposited on the silicon chip and 6 nL of an electrospray buffer solution (0.2% formic acid solution in ACN) was added to one of the samples. Subsequently, this sample was aspirated into a borosilicate glass needle for nESI-MS analysis. Figure 7(a) shows the resulting spectrum. A clear charge envelope signal of the multiply charged protein is observed. To the second deposited volume of the cytochrome C solution, ca. 0.5 nL of trypsin solution (50 µM in water) was added. After 2.5 h, the digestion was terminated by adding 6 nL of the acidic electrospray buffer solution. The solution of the digested cytochrome C was then aspirated into the borosilicate needle and analyzed. The resulting nESI-MS spectrum is shown in Fig. 7(b). Table 1 shows ExPASy PeptideMass database calculations (based on zero missed cleavages) of the generated peptides after the tryptic digestion. The table also shows corresponding masses as well as the amino acid sequence position and amino acid sequence. The peptide numbers are assigned to the peaks shown in Fig. 7(b). The peptide fragments observed in the spectrum based on zero missed cleavages represented an amino acid sequence coverage of 81%. If one missed cleavage is accepted (peaks assigned in parentheses in Fig. 7(b)), an amino acid sequence coverage of 100% was obtained. Thus, the experiment demonstrates that the setup can be utilized to perform chemical reactions in very small sample volumes over an extended period of time, and that the outcome of the chemical reaction can be analyzed with nESI-MS. There is an open access to the sample, before, during and after the reaction, while evaporation of the sample is prevented by the covering of fluorocarbon liquid. This offers a significant degree of freedom. Applications are not limited to reactions with fast kinetics as in the case with most chip-based online systems for nESI-MS analysis. Furthermore, it is possible to select a suitable buffer for the enzymatic reaction and then, by addition of further reagents, change the composition in the samples to facilitate the ESI process.

Figure 4.

Microphotographs of the types of nESI needles utilized throughout the experiments: (a) a quartz needle and (b) a borosilicate glass needle. Enlargements of the tips are shown to the right.

Figure 5.

Mass spectra from analyses of 100 nM of α-lactalbumin (MW 14176 Da) in 49.5:50:0.5 (v/v) water/acetonitrile/formic acid. (a) Analysis using a borosilicate needle. (b) Analysis using a quartz needle.

Figure 6.

Mass spectra of insulin (chain B, oxidized. MW 3496 Da) in 49.5:50:0.5 (v/v) water/acetonitrile/formic acid. (a) Analysis of 800 amol insulin in a 80 nL sample volume, using a quartz needle. (b) Analysis of 105 amol insulin in a 2.1 nL sample volume, using a borosilicate glass needle. (c) Analysis of 105 amol insulin in a 0.5 µL sample volume, loaded with a pipette to a borosilicate glass needle.

Figure 7.

(a) Mass spectrum of 7 µM cytochrome C (MW 12384 Da) in a solution of 70% (v/v) 10 mM ammonium acetate in water and 30% acetonitrile with 0.2% formic acid. (b) Mass spectrum of tryptic digested cytochrome C after 2.5 h of digestion in a sample volume of 15 nL. The electrosprayed solution had a final composition of ca. 7 µM digested cytochrome C in an equivalent sample solution as in (a). The peaks are assigned according to their peptide number with their charge state denoted as the superscripted index. Peaks assigned with series of two peptide numbers in brackets are based on one missed cleavage.

Table 1. The results from an ExPASy database search of tryptic digestion of cytochrome C, used for assignment of the peaks in Fig. 7
Peptide numberMass [M+H]+Amino acid positionAmino acid sequence

The minute sample volumes utilized in the described setup proved to be sufficient for amino acid sequencing with MS/MS. The results are shown in Fig. 8. From a sample volume of 2.2 nL the full amino acid sequence of the 10 amino acids in angiotensin I was obtained. These results demonstrate the potential of the setup, which should be of particular interest for proteomics research.

Figure 8.

MS/MS spectrum of 10 µM angiotensin in 49.5:50:0.5 (v/v) water/acetonitrile/formic acid. The sample volume was 2.2 nL. The peaks are assigned according to their fragmentation pattern with the charge state denoted as the superscripted index.


We have shown that it is feasible to perform nESI-MS analysis from discrete sample volumes down to 1.5 nL, resulting in low attomole absolute detection limits. In a comparative test, using such amounts of analyte in a conventional (0.5 µL) volume, no signal could be detected. Thus, we propose that the confinement of the analyte into a small sample volume is beneficial for obtaining an increased sensitivity. We found that a higher S/N ratio is obtained for equal concentrations of analyte when using needles fabricated in quartz compared to when using borosilicate glass needles. To sustain electrospray and data acquisition from small sample volumes, the flow rate of the generated electrospray should be low, needles with small orifice dimensions and/or flow restrictions such as long, tapered nESI needles are required. Although we have not investigated this systematically, the risk for needle clogging should be considerably less when spraying nanoliter-sized samples compared to when spraying microliter-sized volumes.

In the described setup, the void volume of the transfer capillary to the microchip is ca. 0.5 µL and it should be straightforward to transfer the nanoliter-sized sample volumes from bulk volumes of only a few microliters. It should also be possible to concentrate microliter volumes down to a few nanoliters on the silicon chip by means of controlled evaporation, prior to covering the concentrate with the fluorocarbon liquid. We also consider the possibility to transfer small volumes, obtained from a separation column or from a gel extract, to the surface of the liquid fluorocarbon-covered microchip.

We found that the generated electrospray signal from a 2.2-nL-sized sample volume was sufficient for amino acid sequencing with MS/MS. When performing chemical reactions in the small sample volumes, the chip-based setup with the cover of fluorocarbon liquid provides extended degrees of freedom in terms of addition of multiple reagent solutions into one and the same sample volume as well as the possibility for prolonged reaction times. A challenging task will be to prepare concentrates of analytes of low concentrations from a complex biological matrix into a small volume for subsequent nESI-MS analysis, e.g. by exploiting selective enrichment based on immunoaffinity. We are currently investigating such a strategy.


We thank the Nanochemistry Program, funded by the Swedish Foundation for Strategic Research and the Swedish Research Council, for financial support.