Determination of Melamine and its Analogues from Powdered Infant Milk Using Polymeric Solid-Phase Extraction (SPE) and Reversed-Phase Chromatography

Since the infant milk scandal in 2007, several methods have been developed for the analysis of melamine in dairy products. However, a comprehensive method for the analysis of the related compounds (ammeline, ammelide, and cyanuric acid [see Figure 1]) is difficult to find. The U.S. FDA has such a method, but the lengthy sample preparation steps may not be practical in a production laboratory in which sample preparation time is critical. The hydrophilic nature of the analytes suggests that a hydrophilic interaction liquid chromatography (HILIC) column would be the most promising, and most procedures use a HILIC method. However, HILIC has many disadvantages regarding reproducibility and high organic solvent usage. An alternative method employing Captiva™ ND plates (Agilent Technologies, Lake Forest, CA) provided rapid and reliable results for melamine. Its analogues, on the other hand, were only quantifiable at higher concentrations. A two-tiered approach was thus developed, allowing for the quantification of melamine and its associated analytes at lower concentrations. This method maintains the use of reversed-phase chromatography. Good linearity was achieved for each analyte, as well as high accuracy and precision.

Figure 1 – Structures of melamine, ammeline, ammelide, and cyanuric acid, respectively.


Standard preparation

All compounds were obtained from ChromaDex® (Irvine, CA):

  • Melamine ASB-00013163-100
  • Cyanuric acid ASB-00003958-100
  • Ammeline ASB-00001657-100
  • Ammelide ASB-00001659-100

Melamine and cyanuric acid were prepared in water. Ammeline and ammelide were prepared in a 2% ammonium hydroxide solution. All four compounds were mixed together to a concentration of 100 μg/mL in 2% ammonium hydroxide. Ammonium hydroxide is required to keep all compounds in solution.

Calibration curve

Powdered infant milk was prefortified to concentrations ranging from 0.1 μg/g to 10 μg/g of melamine and its analogues.

Sample preparation

  1. Weigh out 1 ± 0.01 g powdered infant milk
  2. Spike milk with analytes to 1 μg/g
  3. Add 20 mL H2O
  4. Vortex or shake; there should be no remaining powder
  5. Transfer two 1-mL aliquots of milk to test tubes
  6. To one sample, add 2 mL 0.1N HCL (will be applied to Plexa PCX cartridge [Agilent])
  7. To the other sample add 2 mL 0.1N sodium hydroxide (will be applied to Plexa PAX cartridge [Agilent])
  8. Let sit for 30 min


Plexa PCX

1. Conditions
a. 3 mL MeOH
b. 3 mL H2O

2. Loading—apply sample pretreated with 0.1N HCL

3. Wash
a. 3 mL 2% formic acid in H2O
b. 3 mL 50:50 MeOH:ACN

4. Elute—5% ammonium hydroxide in 50:50 MeOH:ACN

Plexa PAX

1. Conditions
a. 3 mL MeOH
b. 3 mL H2O

2. Loading—apply sample pretreated with 0.1N NaOH

3. Wash
a. 3 mL H2O
b. 3 mL MeOH

4. Elute—5% acetic acid in MeOH

Evaporate to dryness and reconstitute in 500 μL 50:50 MeOH:ACN.

Figure 2 – Gradient conditions—flow: 0.4 mL/min; injection vol.: 5 μL; temp: ambient; run time: 3 min.

Instrumentation and conditions

  1. System—Agilent 1290 Infinity LC and Agilent 6460 Triple Quadrupole LC/MS
  2. Column— Agilent Pursuit XRs Ultra 2.8 diphenyl 100 × 2.0 mm PN A7521100X020
  3. Mobile phase—A: 0.1% formic acid in H2O. B: MeOH
  4. Gradient (see Figure 2):

              Time %b
              0.00    2
              0.50    5
              2.00    5
              3.00    80
              3.01    2

      Gas temp: 300 °C
      Gas flow: 5 mL/min
      Nebulizer: 20 psi
      Sheath gas temp: 275 °C
      Sheath gas flow: 7 mL/min
      Capillary: +3500–2000
      Nozzle: +0–500
      Electron multiplier voltage: +300
      Dwell: 300
      Cell accelerator voltage: 7
      Ions: Collected via multiple reaction monitoring (MRM) see (Table 1)

Table 1 – Multiple reaction monitoring ions

Results and discussion

Although this method requires separate preparation for cyanuric acid, a single reversed-phase LC run can be used for all compounds. The samples treated with HCL were applied to Plexa PCX cartridges. From those cartridges melamine, ammeline, and ammelide were extracted. The cyanuric acid samples were pretreated with NaOH and extracted on the Plexa PAX cartridges.

Figure 3 – Separation of a 10-ng/mL standard of melamine and its analogues.

Figure 3 demonstrates the baseline separation of the melamine–cyanuric acid pair and the ammeline–ammelide pair. All compounds were reliably detected down to 0.2 μg/g.

Figure 4a – Calibration curves of melamine.
Figure 4b – Calibration curves of cyanuric acid.
Figure 4c – Calibration curves of ammeline.
Figure 4d – Calibration curves of ammelide.

The compounds showed linearity from 0.5 μg/g to 10 μg/g with R2 above 0.993. A minimum of five levels were used for the calibration curves (Figure 4).

Table 2 – Recoveries of melamine and its analogues from fortified powdered infant formula compared to Captiva method, 1.0 μg/g (n = 6)

Table 2 lists the relative recoveries of melamine and its analogues following sample preparation with Plexa PCX and Plexa PAX. All analytes extracted with Plexa PCX showed relative recoveries within 14% of true value with RSDs below 8%. Cyanuric acid, extracted from Plexa PAX, showed relative recovery within 17% of true value with RSD of less than 6%.

Figure 5 – Melamine extracted through Plexa PCX (top) and Captiva ND (bottom) at 1.0 μg/g.

Analyte sensitivity increased for all compounds. Figure 5 demonstrates a threefold sensitivity increase for melamine compared to a similar sample prepared using Captiva ND. Both were prefortified to 1.0 μg/g.


This article describes a method for the rapid analysis of melamine and its analogues from powdered dairy products using generic sample preparation procedures for Plexa PCX and Plexa PAX and reversed-phase chromatography. Analysis of melamine and its analogues was accomplished down to the regulatory levels of 1 μg/g1 and below. Linearity was demonstrated for all compounds with a linear regression coefficient greater than 0.993.

The previous Captiva ND method was sensitive enough for melamine only, maintaining a signal-to-noise ratio (SNR) >5:1 as required by the U.S. FDA at the 1-μg/g level.2 The current method is sensitive for melamine and all of its analogues below the regulatory level of 1 μg/g. Ammeline, ammelide, and cyanuric acid maintained an SNR above 5:1 down to the 0.1-μg/g level. Melamine was capable of maintaining that SNR down to 0.5 μg/g. At regulatory levels, melamine, ammeline, and ammelide showed relative recoveries within 14% of true value with RSDs below 8%. Cyanuric acid demonstrated a relative recovery within 17% of true value with RSD of less than 6%.

Use of a two-tiered approach with Plexa PCX and Plexa PAX for the analysis of melamine and its analogues allows for increased sensitivity without sacrificing time or ease of use.


  1. Chinese Ministry of Health (2011). China sets limits of melamine levels tolerable in food products, 21 April, 2011;
  2. Turnipseed, S.B.; Casey, C. et al. U.S. FDA Laboratory Information Bulletin 2008, 24: 4421. Lib. no. 4421;; accessed Sept 1, 2011.

Daniel Tran is Research Chemist, Agilent Technologies, 25200 Commerce Dr., Lake Forest, CA 92630, U.S.A.; tel.: 949-770-9381; e-mail: