Reply to Soper: Density measurement of confined water with neutron scattering

LETTER

Reply to Soper: Density measurement

of con

fined water with neutron

scattering

This is a response to Soper’s two comments (1) regarding our papers (2, 3) in PNAS that (a) the distribution of water across the pores is not uniform and (b) the majority of water may reside outside the pores. Here, we show that we have given proper consideration to both issues and have reconfirmed the validity of our method and conclusion as elaborated in the following.

The possibility that layering effects across the pores may in-troduce errors in associating the (100) interchannel peak height with density is not a new idea (reference 3 in ref. 1), and it has already been addressed (2). The arguments of Soper (4) mainly rest on the assumption that the average density of water does not depend on temperature, which is drawn by linking the temperature-independent incoherent background of his ex-perimental data to the water density and not to the amount of sample exposed to the neutron beam. As a result, he constructs a density profile with a large spatial variation across the pores in an effectively underfilled condition ðρwater¼ 0:70 g=cm3Þ (4), reproduced as Fig. 1A. The existence of such voids is disproved by the measurement on the contrast-matched sample with D2O/

H2O mixtures, because no evidence of the scattering from the

voids can be recognized. Although the water distributions sug-gested by Soper (4) may reproduce the experimentally measured intensity change of the (100) interchannel peak at some selected temperatures, they are inconsistent with numerous experi-mental and simulation studies (references 32 and 55–61 in ref. 2) and introduce unnecessary complications to the model. On the other hand, we have estimated the effect of a density profile based on these published simulation results (Fig. 1 B–D). The layering effect on the (100) peak intensity is found to be negli-gible compared with the experimentally observed intensity change shown (figure 2 in ref. 2) as long as the spatial variations do not exceed such a level. We therefore believe our average density approach to be valid and accurate.

As for the second point raised by Soper (4), we have confirmed that the amount of excess water is negligible compared with the

interior water by measuring the water vapor adsorption-de-sorption isotherm. Also, both differential scanning calorimetry scans and inelastic neutron scattering measurements of the generalized librational density of state do not detect any signs of freezing. In any case, a small amount of external water is unlikely to affect the behavior of water within the pores. The (100) in-terchannel peak is sensitive only to the scattering contrast be-tween the water in the channels and the surrounding silica and has nothing to do with the excess water. We are aware that the pore size, especially for those<20 Å, may be considerably un-derestimated by the standard Barrett–Joyner–Halenda (BJH) method. The BJH pore size of 15 Å used in our paper is thus a nominal value to allow meaningful comparisons with other measurements. Using the Kruk–Jaroniec method (5), the pore size of the MCM-41-S-15 sample may well be 25 Å, implying a hydration level of 0.45 gD2O/gSiO2using Soper’s formula (4),

which is close to our measured value.

Yang Zhanga, Antonio Faraoneb,c, William A. Kamitakaharab, Kao-Hsiang Liud_{, Chung-Yuan Mou}d_{, Juscelino B. Lea˜o}b_{, Sung Chang}b_,

and Sow-Hsin Chene,1

a_{Neutron Sciences Directorate and Joint Institute for Neutron}

Sci-ences, Oak Ridge National Laboratory, Oak Ridge, TN 37831;

b_{NIST Center for Neutron Research, National Institute of}

Stand-ards and Technology, Gaithersburg, MD 20899; cDepartment of Materials Science and Engineering, University of Maryland, College Park, MD 20742; dDepartment of Chemistry, National Taiwan University, Taipei 106, Taiwan; and e_{Department of Nuclear}

Sci-ence and Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139

1. Soper AK (2011) Density minimum in supercooled confined water. Proc Natl Acad Sci USA 108:E1192.

2. Zhang Y, et al. (2011) Density hysteresis of heavy water confined in a nanoporous silica matrix. Proc Natl Acad Sci USA 108:12206–12211.

3. Liu D, et al. (2007) Observation of the density minimum in deeply supercooled confined water. Proc Natl Acad Sci USA 104:9570–9574.

4. Soper AK (2011) Density profile of water confined in cylindrical pores in MCM-41 silica. ArXiv e-prints. Available at http://arxiv.org/abs/1107.3492v2. Accessed July 20, 2011. 5. Kruk M, Jaroniec M, Sayari A (1999) Relations between pore structure parameters and

their implications for characterization of MCM-41 using gas adsorption and X-ray diffraction. Chem Mater 11:492–500.

Author contributions: Y.Z., A.F., W.A.K., K.-H.L., C.-Y.M., and S.-H.C. designed research; Y.Z., A.F., W.A.K., K.-H.L., C.-Y.M., J.B.L., S.C., and S.-H.C. performed research; and Y.Z., A.F., W.A.K., K.-H.L., C.-Y.M., and S.-H.C. wrote the paper.

The authors declare no conflict of interest.

1_{To whom correspondence should be addressed. E-mail: sowhsin@mit.edu.}

A

B

C

D

Fig. 1. Comparison of the density profiles and their simulated intensities. (A) Density profiles constructed by Soper (4). Note that water is allowed to penetrate into the MCM-41-S silica material. The simulated intensity can be found infigure 7 in ref. 4. (B) Density profile based on published simulation results with a 3-Å surface water layer of 10% higher density, for which the computed intensity, I(Q), and particle structure factor, P(Q), are presented in C and D with two different approaches: the average density approach (green) and the core-shell model (red). The simulated intensities with the two approaches show negligible difference at the (100) interchannel peak compared with the experimentally observed intensity change shown infigure 2 in ref. 2.