By: Rich Sportsman, President (March 27, 2020)
There are several reports comparing various methods for determining free and total sulfites (SO2). Aeration oxidation (AO) and Ripper titration methods (like Vinmetrica, Hanna, and old starch color change techniques) are the most commonly employed. In addition there are a number of methods based on spectrophotometric techniques or color reactions that have come into use in recent years. Here I want to focus on the performance of the AO and Ripper methods vs. others, and vs. each other.
We have generated lots of our own data comparing AO and Vinmetrica’s SO2 test, and we know without any doubt that, with the exception of exotic samples (like wines made from ascorbic acid-containing fruit), the Vinmetrica system produces results that are indistinguishable from those of the AO method. But I think it’s time to revisit the information that’s out there on others’ assessment of the comparison of these techniques.
An independent evaluation
First I want to bring up the work done by Daniel Pambianchi whose web site Techniques in Home Winemaking http://techniquesinhomewinemaking.com/ provides very useful resources for devotees of home winemaking. Daniel likes to include reviews of wine analysis products among the many offerings on his web pages. But his reviews are not just impressions; he likes to get into his lab and put things to the test. A few years back, he took a look at measuring free SO2 with various methods. You can find the article here: http://techniquesinhomewinemaking.com/attachments/File/Benchmarking%20of%20SO2%20Analysis%20Instruments%20and%20Methods.pdf
Figure 1 of Daniel’s article, shown above, shows that for tests done on a simple water standard with 35 mg/L free SO2, two AO systems and the Ripper-based Vinmetrica and Hanna systems come within 1.5 mg/L of the expected value. The precision of the first three looks good while that of the Hanna system appears rather poor. In other parts of the article, Daniel looks at interference from tannins and ascorbic acid, and shows that, as expected, the former don’t have an effect on either technique outside of the expected binding of free SO2, while the latter interferes with the Ripper methods but not with AO.
There is an organization known as Collaborative Testing Systems (CTS) in Virginia that provides assessment of the performance of testing laboratories in various disciplines. One of these programs is in the area of wine testing (https://collaborative-testing.com/program-8.php). CTS provides standard wine samples to participants and then tabulates and analyzes the results to assess inter-laboratory agreement on results for SO2, ABV, residual sugar, and many other tests.
From their web site, you can view the data assembled for free SO2, which gives a glimpse of the performance of these tests by a variety of methods from labs far and wide. I refer below to two such data sets:
In this report from 2019, 81 of 83 participants reported usable free SO2 data on two chardonnays provided by the CTS program. Of these, 26 reported AO results and 29 reported Ripper results. The “grand mean” (g.m.), i.e., the average of results by all methods, for chardonnay SA25 was 24.1 mg/L with a standard deviation (±) of 3.0, while the grand mean of SA26 was 25.4 mg/L ± 4.4.
For SA25, Ripper methods averaged 24.8 ± 2.7 while AO was 22.8 ± 3.1 [g.m. 24.1 ± 3.0]
For SA26, Ripper methods averaged 26.0 ± 3.8 while AO was 25.1 ± 5.2 [g.m. 25.4 ± 4.4]
Zinfandel and Red Blend
In this report also from 2019, 95 of 103 participants provided usable free SO2 data on a zinfandel and on a red blend. Of these, 34 reported AO results and 28 reported Ripper results. The grand mean for Red Blend SA23 was 24.0 mg/L ± 2.7, while the grand mean of zinfandel SA24 was 22.4 mg/L ± 2.8.
For SA23, Ripper methods averaged 24.6 ± 2.5 while AO was 23.5 ± 2.6 [g.m. 24.0 ± 2.7]
For SA24, Ripper methods averaged 23.1 ± 2.7 while AO was 22.2 ± 2.7 [g.m. 22.4 ± 2.8]
I think these results pretty clearly lay to rest any concern that Ripper titrations produce free SO2 results that are different from those of AO methods. In all cases above, most average results by both methods are within 1 mg/L of the grand mean (AO missed by a little more on the SA25 chardonnay). In addition, the standard deviations of Ripper averages are less than those of the grand means, while the same is not true for all the AO results.
Now, we don’t know how many of the Ripper results were done using a Vinmetrica system, and we have not ourselves yet participated in the CTS program (though we plan to do so this year). But the main alternatives to the Vinmetrica system for Ripper titrations are the Hanna autotitrators and the use of starch indicator. Since all use similar chemistry, the only difference between these methods is the detection of the endpoint, assuming everyone has made and used their solutions correctly.
We won’t discuss the starch endpoint method of Ripper, which is pretty much impossible to use in a red wine. The Hanna autotitrator uses an ORP electrode to detect the endpoint. The change in response of an ORP probe in this system is rather subtle and requires data analysis to call it – below is shown data from Thermo’s web site using a similar set-up but manually titrating.
http://www.qclscientific.com/pdfs/Apps Notes/Sulfur Dioxide in Wine.pdf
To determine the actual endpoint, it is necessary to titrate past the endpoint to be able to tell where it occurred by plotting something like change in mV per drop of titrant. In the above figure, the endpoint occurs at about 3.6 mL of titrant, but the titration needs to go out to about 4 mL to collect enough data to make this assessment.
By contrast, the Vinmetrica system detects the endpoint by responding to the first appearance of free iodine after all the SO2 has reacted with the Titrant. As shown below, the endpoint is given by the first increment of Titrant that causes a significant rise in the electrochemical current, in this case, when 1.9 mL was added. (One drop less – 1.85 mL added – the current was below 20 nA and so not called) Going past the endpoint causes further increases in this current, but this information is not needed to call the endpoint, which is sharp and clear.
A final point is made on this slide from a presentation at the Unified Symposium last year. This compares over 3400 total SO2 results reported to CTS over 13 years up to 2015, breaking them out by technique. Based on the appearance and degree of normal distribution about the grand mean, CTS concludes that Ripper and AO have preferred performance compared to all other methods used.
So in conclusion, there’s no reason to assume a Ripper titration is somehow a compromise between accuracy and simplicity. You get both, especially if you use the Vinmetrica system.