Integration of a Dilution Module in a Mass Spectrometry-Based Online Reaction Monitoring System

Featured Article

 Integration of a Dilution Module in a Mass Spectrometry-Based Online Reaction Monitoring System

Online reaction monitoring enables the study of reaction mechanisms and reaction kinetics through the monitoring of reactants, products and transient species, and can be used for real-time data acquisition1 without unnecessary interruptions or sample manipulation. Among the spectroscopic techniques used for online reaction monitoring, i.e., fluorescence, UV/VIS, IR, Raman, X-ray and nuclear magnetic resonance (NMR) spectroscopy, mass spectrometry offers specific information regarding molecular mass and the structure of components in a reaction mixture with higher accuracy, sensitivity and speed.2 However, due to the high molar concentration of reaction mixtures, sample overloading can occur, resulting in excitation of the detector operating range. Sample overload can also reduce lifetime and result in distorted mass spectra.3,4

One solution is to modify the ionization technique. Ionization techniques such as low-temperature plasma (LTP), extracted electrospray ionization (EESI) and Direct Analysis in Real Time (DART, marketed by JEOL [Peabody, Mass.] and IonSense [Saugus, Mass.]) analyze chemical reactions at molar concentrations.5 Another approach is dilution of the analytical solution followed by thorough mixing.4 Dilution can be based on the ratio of input flows such as varying the time,6 flow rate,7 frequency8 or volume.9

Clinton et al. used a microporous membrane interface to achieve high dilution ratios ([initial volume]/[final volume]).10 A donor solvent containing the sample was pumped to waste through the groove on one side of the membrane interface. Acceptor solvent was pumped through the other side of the membrane in the opposite direction into an atmospheric pressure chemical ionization-quadrupole mass spectrometer (APCI-MS). A dilution ratio of approx. five orders of magnitude was achieved for the Michael addition reaction of phenylethylamine (PEA) and acrylonitrile in ethanol. The analytical response was influenced by changes in pressure and flow rate of the donor and acceptor solvents.5

Dell’Orco et al. used HPLC pumps and a passive flow splitter to dilute the reaction mixture (piperidine-catalyzed Knoevenagel reaction) by approx. 3000 times. The important reaction intermediates and reaction kinetic time scale were identified, but quantitation of these species was still problematic.3 Using an HPLC pump and a flow splitter coupled with a flow injection analysis/atmospheric pressure chemical ionization (FIA/APCI) MS, Zhu et al. reported quantitative real-time monitoring of a model reaction performed at 1.63 mol/L, a concentration often required in process control.5 Brum et al. and Browne et al. added a pump to the main stream via a mixing tee for low ratio dilution.11,12

For dilutions up to 100, 200, 500 or even 1000, another method includes use of an active splitter or mass rate attenuator (MRA), which can reach a dilution ratio of up to 100,000. Cai et al.4 replaced a passive splitting device with an MRA in a purification system, which resulted in improved system plumbing, volume reduction of the collected fraction and reduced evaporation time to reclaim purified products. Toribio and Leister13,14 employed MRA in the purification of combinatorial libraries, drug metabolites or characterizable impurities. MRA was used for sampling and transferring the analyte(s) to the dilution flow by Bristow et al.1 This make-up flow was then sent to a portable, low-footprint mass spectrometer. Jeurissen et al.15 configured an MRA after the HPLC column for diluting extracts in an online HPLC detection system. The main usage of the MRA in this application was to lower the solution concentration to a suitable working range of the MS.

Reaction monitoring system

For the current study, a mobile reaction system based on FIA was coupled to an electrospray ionization/time-of-flight (ESI/TOF) MS.16 Analytical solution from the microreactor outflow was injected after dilution into a carrier stream, which was coupled to the analyzer. The dilution module consisted of an MRA, dilution pump and compensation pump, and was able to perform dilutions at different ratios. Settings were determined automatically based on the desired dilution ratio and flow rate in the microreactor, which varies according to the reaction stages.


WellChrom K-501 and Smartline 100 HPLC pumps (Knauer GmbH, Berlin, Germany) were used to transfer reactants/educts to the 11.2-mL meander reactor (Ehrfeld Mikrotechnik BTS, Wendelsheim, Germany). The reactor solution temperature was controlled by the F26-HP refrigerated/heating circulator (Julabo, Allentown, Penn.). After exiting the microreactor, the solution was diluted at the dilution module, and then entered a 10-port switching valve via the two-position microelectric valve actuator (Valco Instruments, Houston, Texas). The solution was injected into the carrier stream controlled by the mzr-2942 microannular gear pump (HNP Mikrosystem GmbH, Schwerin, Germany) and CORI-FLOW model M13 mass flowmeter (Bronkhorst Mättig, Kamen, Germany). Finally, the sample was transferred to the TOF-MS G1969A (Agilent Technologies, Santa Clara, Calif.) for analysis.16,17

Dilution module

The dilution module (Figure 1) comprised the MRA-100-00 (Rheodyne, L.P., Rohnert Park, Calif.) and WellChrom K-501 HPLC pump. Output from the microreactor flowed through the MRA, and then a portion of the output stream split off and entered the dilution flow. The amount of output stream sent to the dilution flow was determined by the split factor set on the device, which governs the switching frequency (0.2 Hz–2 Hz) and aliquot volume (22, 100 and 300 nL). Dilution ratios up to 100,000 are possible by changing the split factor and dilution flow rate.

Figure 1 – Top: Schematic view of system and dilution module. Bottom: System: 1) educt pumps, 2) microreactor, 3) MRA, 4) multiport valve, 5) solvent carrier module (microannular gear pump and mass flow controller), 6) dilution pump, 7) compensation pump, 8) mass analyzer, 9) heating/cooling module, 10) communication module.

Because microreactor output can be varied to monitor different reaction stages, input to the MRA can be altered as well. For stable dilutions, the split amount should be stable to different input flow conditions. However, initial testing showed strong dependence of the signal intensity (peak area) of the diluted solution on the microreactor flow rate, especially below 3 mL/min (Figure 2a). This unexpected influence of flow rate variation on the splitter impacts the dilution. The signal intensity of the diluted flow shows higher stability with microreactor flow rates greater than 3 mL/min; thus an additional pump (G1310A, Agilent Technologies) was added. This isocratic pump is activated when the flow does not reach 3 mL/min. In turn, the dilution flow rate is reduced, corresponding to the added portion to the main stream. This dilution process is automatic so that the split factor, dilution flow rate and compensation flow rate are automatically selected based on the dilution ratio and reaction stages set by the user. For example, with a dilution ratio of 1:100, if the microreactor flow rate is 4 mL/min (>3 mL/min), a split factor of 1 and dilution flow rate of 1 mL/min are selected. When the microreactor flow rate is 1 mL/min (<3 mL/min), the compensation flow is 3 – 1 = 2 mL/min. In this case, the original solution is diluted by 66.7%, the required dilution ratio now being only 1:33. Therefore, the dilution flow is reduced to 0.33 mL/min to achieve the desired dilution ratio (1:100). Results for the modified dilution approach are presented in Figure 2b, which shows improved stability from flow rate fluctuation of the microreactor.

Figure 2 – a) Dependence of diluted solution on flow rate of microreactor. Split factor (3) and dilution flow rate (0.6 mL/min) were fixed. Only the microreactor flow rate (or sample flow rate) was varied from 0.6 mL/min to 6 mL/min. Test compounds: OS and 4-MOS in MeOH. b) Dilution with flow compensation approach with the microreactor flow rate from 0.6 mL/min to 6 mL/min.


Materials and methods

Acetonitrile ROTISOLV and methanol (both HPLC gradient grade) were purchased from Roth (Karlsruhe, Germany); octanoic acid (OS) (>99%) was obtained from Merck-Schuchardt (Hohenbrunn, Germany) and 4-methyloctanoic acid (4-MOS) (>98%) was from Sigma-Aldrich (a part of MilliporeSigma, Darmstadt, Germany). Concentrations of OS and 4-MOS in MeOH solvent were prepared as follows: stock solution 1 (low concentration) contained 91 mg/L OS and 45.5 mg/L 4-MOS; stock solution 2 (medium concentration) 182 mg/L OS and 91 mg/L 4-MOS and stock solution 3 (high concentration) 273 mg/L OS and 136.5 mg/L 4-MOS.

For every concentration of the stock solution, four dilution ratios were performed in descending order—1:500, 1:300, 1:100 and 1:50. With every dilution ratio, different reaction stages of 3, 6 and 12 min were selected, which corresponded to microreactor flow rates of 4, 2 and 1 mL/min. These inputs determined the split factor, compensation flow rate and dilution flow rate. From the stock solutions, manual dilutions were also performed at the same dilution ratios in order to compare the results obtained by automatic dilution.

The MS was operated in negative ion polarity mode, with scanning rate of 1 spectrum/sec and mass range of 100–500 m/z. Nitrogen with a flow rate of 10 L/min was used as the drying gas. Source temperature was at 350 °C and nebulizer pressure was 40 psig. The following voltages were applied: capillary 3500 V, fragmentor 175 V and skimmer 65 V.

System settings

A mixture of ACN/H2O (40/60%) was used as the carrier solvent stream to the MS. Methanol was used as diluent by the dilution pump and compensation pump. Other settings were flow rate ratio of the educt pumps 1:1, total reaction volume 12.18 mL, carrier solvent flow rate 0.6 mL/min, sampling time 0.2 sec, sampling interval 18 sec and sampling repetition 7. Each dilution was repeated at least twice.

Results and discussion

Dilution results with stock solution 2 (182 mg/L OS + 91 mg/L 4-MOS in MeOH) are shown in Figure 3 and in Table 1 for OS (similar behavior was observed for 4-MOS; thus only the results of OS are presented). In every dilution ratio, the average peak area of each reaction stage (3, 6 and 12 min) was calculated. The average peak area of all peaks for every dilution ratio and standard deviation was also determined. These results show a logical increase of the sampling peak areas (from 549,566 to 5,993,738 for OS and 356,885 to 4,196,049 for 4-MOS) as the dilution ratio decreases (from 1:500 to 1:50). They are also equivalent to the manually diluted solutions of similar ratios. Manual dilution is not affected by the microreactor flow rate variation in the testing range because the sampling loop is only 5 μL (see Table 1). Due to the (possibly) different settings of the dilution module (for split factor, compensation flow rate and dilution flow rate) and microreactor flow rates for each reaction stage, the standard deviations measured with automatic dilution (4.3–14.4% for OS) are normally higher and more wide-ranging than those with manual dilution (7.5–10.7%). Further comparison of the signal intensity between different dilution ratios (1:50, 1:100, 1:300) with a dilution ratio of 1:500 yielded values in two rows—expected quotient and actual quotient. It is expected that the signal intensity of dilution ratio 1:50 would be 10 times that with a dilution ratio of 1:500. The actual quotient (5,993,738/549,566) was 10.91, which is acceptable. Other logical quotients were obtained, i.e., expected quotient: 5 – actual quotient: 5.77 (3,168,592/549,566) with dilution ratio 1:100 and with dilution ratio 1:300—expected quotient: 1.67 – actual quotient: 1.93 (1,062,333/549,566).

 Figure 3 – Automatic and manual dilution with ratios of 1:500, 1:300, 1:100 and 1:50 of OS in MeOH.
Table 1 – Automatic and manual dilution of OS in MeOH (run 2; solution: 182 mg/L OS + 91 mg/L 4-MOS in MeOH)

Figure 4 shows the results of automatic dilution with the same dilution ratios but different concentrations of stock solution (solutions 1, 2 and 3). Only the signal of octanoic acid is plotted. As can be seen, for every concentration of the stock solution, the signal intensity of the diluted solution increases as the dilution ratio decreases from 1:500 to 1:50. In addition, for each of the same dilution ratios, the peak areas of stock solution with higher concentration are higher. This results in a much steeper slope of the green curve (from stock solution 3), red curve (from stock solution 2) and blue curve (from stock solution 1).

 Figure 4 – Dependence of diluted solution on stock solution concentration.

Similar tests were performed using different solvents, such as ACN/H2O (40%/60%) and pure ACN. For each solvent, similar stock concentrations were prepared (solution 1: 91 mg/L OS + 45.5 mg/L 4-MOS; solution 2: 182 mg/L OS + 91 mg/L 4-MOS; solution 3: 273 mg/L OS + 136.5 mg/L 4-MOS). ACN/H2O (40%/60%) and pure ACN were used as diluents by both the dilution pump and compensation pump. Figure 5 shows the plots for 182 mg/L octanoic acid in different solvents (MeOH, ACN, ACN/H2O). The effect of different solvents (with characteristics such as specific conductance, surface tension and viscosity) on the ionization efficiency of analytes is not remarkable since the three curves are not significantly separated. As expected, the dilution module can dilute different solvents. Decreasing the dilution ratio from 1:500 to 1:50 increases the peak areas of the diluted solution. In addition, there is an increase in peak area at each dilution ratio as the concentration of stock solution increases, which results in different slopes of the diluted solutions (not shown).

 Figure 5 – Automatic dilution with different stock solution solvents.

It is possible to reduce the dilution flow rate to achieve dilution ratios such as 1:2 or 1:5. However, laboratory experiments showed a signal reduction when the dilution flow was below 0.3 mL/min. Therefore, the lowest dilution ratio that can be achieved with this method is around 1:10.


A dilution module based primarily on a mass rate attenuator solves the problem of high concentration solution to the mass spectrometer. The low-flow compensation approach was used to circumvent the strong dependence of diluted solutions on the low flow rate of the microreactor. Although there are high standard deviations due to the effects from different flow rates and device performance, the process is automated and able to perform dilution ratios starting from 10.


  1. Bristow, T.W.T.; Ray, A.D. et al. J. Am. Soc. Mass Spectrom. 2014, 25(10), 1794–1802.
  2. Fabris D. Mass Spectrom. Rev. 2005, 24(1), 30–54.
  3. Dell’Orco, P.; Brum, J. et al. Anal. Chem. 1999, 71(22), 5165–70.
  4. Cai, H.; Kiplinger, J.P. et al. Rapid Commun. Mass Spectrom. 2002, 16(6), 544–54.
  5. Zhu, Z.; Bartmess, J.E. et al. Anal. Chem. 2012, 84(17), 7547–54.
  6. Ainla, A.; Gözen, I. et al. Anal. Chem. 2009, 81(13), 5549–56.
  7. Dertinger, S.K.W.; Chiu, D.T. et al. Anal. Chem. 2001, 73(6), 1240–6.
  8. Zhang, X.; Roper, M.G. Anal. Chem. 2009, 81(3), 1162–8.
  9. Paegel, B.M.; Grover, W.H. et al. Anal. Chem. 2006, 78(21), 7522–7.
  10. Clinton, R.; Creaser, C.S. et al. Anal. Chim. Acta. 2005, 539(1–2), 133–40.
  11. Brum, J.; Dell’Orco, P. et al. Rapid Commun. Mass Spectrom. 1998, 12(11), 741–5.
  12. Browne, D.L.; Wright, S. et al. Rapid Commun. Mass Spectrom. 2012, 26(17), 1999–2010.
  13. Toribio, A.; Destandau, E. et al. Rapid Commun. Mass Spectrom. 2009, 23(12), 1863–70.
  14. Leister, W.; Strauss, K. et al. J. Comb. Chem. 2003, 5(3), 322–9.
  15. Jeurissen, S.M.F.; Claassen, F.W. et al. J. Chromatogr. A. 2007, 1141(1), 81–9.
  16. Fleischer, H.; Hoffmann, D. et al. IEEE I2MTC 2015, 1561–6.
  17. Do, V.Q.; Hoffmann, D. et al. PET ACECS 2016, 528–34.

Vinh Quang Do and Kerstin Thurow are with the Center for Life Science Automation (celisca), Friedrich-Barnewitz-Str., 818119 Rostock, Germany; tel.: +49 (381) 498-7810; e-mail: [email protected]; Heidi Fleischer and Dany Hoffmann are with the Institute of Automation, University of Rostock, Germany. The authors wish to thank the Vietnam Ministry of Education and Training for the Project 911 Scholarship for financial support of this study, and the Federal Ministry of Education and Research Germany (FKZ: 03Z1KN11).

Related Products