Changing Coefficient Lower Bound While Fitting Significantly Reduces R-Square, Despite The Bound Not Restricting The Fit

I will use the Curve fitting toolbox as an example, because I think it is the easiest way to reproduce this erros. I'm using this set of data:

x: [0.150000000000000;0.199800000000000;0.225000000000000;0.240000000000000;0.250000000000000]

y: [0.005516000000000;0.005363000000000;0.005293000000000;0.005257000000000;0.005235000000000]

And I am trying to fit this data with a polynomial fit, of degree two. I am not using the robust option and the center and scale box is unchecked. I know that my data should always be rising on the left side. To set this behavior I am setting the lower bound of the p1 Coefficient to 0, as shown below.

The problem is that the fit is changing, and getting worse when I set this lower bound, despite the fit before setting the lower bound already having a positive p1 value. Here are the results from before and after setting the bound, respectively.

My question is how can I improve this fit while also mantaining the p1 restriction? This can be via code or through the Curve Fitting Toolbox, It will eventually converted to code later anyways.